From jeff at jumpstartlab.com  Wed Mar  2 10:01:10 2011
From: jeff at jumpstartlab.com (Jeff Casimir)
Date: Wed, 2 Mar 2011 07:01:10 -0800
Subject: [raleigh.rb] No West End meetup for March,
	Ruby Primer course instead
In-Reply-To: <AANLkTikQw2qo6i0y5MGqHiy6DJ5GC9qaOq49Pd4atk6h@mail.gmail.com>
References: <AANLkTikQw2qo6i0y5MGqHiy6DJ5GC9qaOq49Pd4atk6h@mail.gmail.com>
Message-ID: <AANLkTim3akzVVgSs7hNkQoCnyRFBYoJv=DqpHTM7HXx=@mail.gmail.com>

Hi all,

We have about three seats left for tonight.  If you'd like to join us,
signup here:

http://jumpstartlab.com/trainings/13-ruby-raleigh

- Jeff

---
Jeff Casimir
Jumpstart Lab by Casimir Creative, LLC
http://jumpstartlab.com
@jumpstartlab on twitter



On Wed, Feb 23, 2011 at 11:04 AM, TJ Stankus <tj at stank.us> wrote:
> This is just a reminder that in lieu of our regular monthly West End
> Ruby meetup, on March 2nd Carrboro Coworking is hosting a FREE Ruby
> Primer Course, taught by Jeff Casimir of Jump Start Lab.
>
> http://jumpstartlab.com/trainings/13-ruby-raleigh
>
> Jeff has a reputation as an awesome teacher. This is a great
> opportunity for anyone who's looking to get going with Ruby. It looks
> like there are some spots left, too!
>
> If you're planning to attend, please note the course starts at 6pm.
>
> We'll resume regular West End meetups in April.
>
> -TJ
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>

From nathaniel at talbott.ws  Wed Mar  2 13:50:55 2011
From: nathaniel at talbott.ws (Nathaniel Talbott)
Date: Wed, 2 Mar 2011 13:50:55 -0500
Subject: [raleigh.rb] Carrburritos at 5
Message-ID: <AANLkTimGZCukbo20wVJW-ZYndGmG53_14SiZmryp-ScJ@mail.gmail.com>

I'm hitting up Carrburritos at 5 tonight before crashing the Ruby
Primer for an hour so. Would love to have company!

http://t.co/GtOidTF


-- 
Nathaniel Talbott
<:((><

From efg at ncsu.edu  Fri Mar  4 11:47:55 2011
From: efg at ncsu.edu (Ed Gehringer)
Date: Fri, 04 Mar 2011 11:47:55 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
Message-ID: <4D70D171.7D16.00A7.1@gw.ncsu.edu>

We are going to use a module to act as an interface, as described in 
http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby 
Is there any convention for naming an interface, e.g.,
 
the interface for a contributor object should be called "contributor-interface"?
 
Thanks,
Ed
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/6288c60f/attachment.html>

From rick.denatale at gmail.com  Fri Mar  4 13:29:03 2011
From: rick.denatale at gmail.com (Rick DeNatale)
Date: Fri, 4 Mar 2011 13:29:03 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
Message-ID: <AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>

IMHO, the right answer to questions like this is MU
http://en.wikipedia.org/wiki/Mu_(negative)

This idea is cargo culting an idea from languages like C++ and Java.
If you want a statically typed language use one, Ruby is perfectly
fine without such stuff.

Of course I might be wrong <G>

On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
> We are going to use a module to act as an interface, as described in
> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby
> Is there any convention for naming an interface, e.g.,
>
> the interface for a contributor object should be called
> "contributor-interface"?
>
> Thanks,
> Ed
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>



-- 
Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Github: http://github.com/rubyredrick
Twitter: @RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale

From mshiltonj at gmail.com  Fri Mar  4 13:44:56 2011
From: mshiltonj at gmail.com (Steven Hilton)
Date: Fri, 4 Mar 2011 13:44:56 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
Message-ID: <AANLkTinST6syTHz-YxEfpbYWGGvLVSM-JfTw2qpXoF1R@mail.gmail.com>

Curious:

Are you going to be able to enforce return values and parameters lists
as well as the method name?

Also, when is interface adherence going to be enforced? During run
time? Why not put such interface enforcement in test suite, out of
production code?

As for naming:
Foo # The Foo Interface
FooImpl # The Implementation of the Foo Interface.

Yo dawg, I herd you like to java, so I put some java in your ruby, so
you can java while you ruby.


On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
> We are going to use a module to act as an interface, as described in
> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby
> Is there any convention for naming an interface, e.g.,
>
> the interface for a contributor object should be called
> "contributor-interface"?
>
> Thanks,
> Ed
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>

From tcrawley at gmail.com  Fri Mar  4 11:04:27 2011
From: tcrawley at gmail.com (Tobias Crawley)
Date: Fri, 4 Mar 2011 11:04:27 -0500
Subject: [raleigh.rb] Interested in a talk on TorqueBox?
Message-ID: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>

I'm looking for opportunities to talk about TorqueBox (http://torquebox.org)
- would you guys be interested?

If you haven't heard of TorqueBox - it's a ruby application server built on
top of JBoss AS (using JRuby), and has integrated messaging, asynchronous
processing, scheduled jobs, clustering, and more.

I'm a member of the TorqueBox team. I live in Asheville, but would be happy
to make the drive down.

Toby
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/d057ea45/attachment.html>

From efg at ncsu.edu  Fri Mar  4 14:08:41 2011
From: efg at ncsu.edu (Ed Gehringer)
Date: Fri, 04 Mar 2011 14:08:41 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
Message-ID: <4D70F26B.7D16.00A7.1@gw.ncsu.edu>

I meant "contributor_interface", of course :-)

>>> "Ed Gehringer" <efg at ncsu.edu> 3/4/2011 11:47 AM >>>

We are going to use a module to act as an interface, as described in 

http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby

Is there any convention for naming an interface, e.g., 

  

the interface for a contributor object should be called "contributor-interface"? 

  

Thanks, 

Ed 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/0e826436/attachment.html>

From efg at ncsu.edu  Fri Mar  4 14:30:44 2011
From: efg at ncsu.edu (Ed Gehringer)
Date: Fri, 04 Mar 2011 14:30:44 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
Message-ID: <4D70F796.7D16.00A7.1@gw.ncsu.edu>

I think you do have it wrong, Rick. 

We have a good application of unbounded polymorphism, where we have a receiver that may be a member of classes that are not in the same inheritance hierarchy.  That's definitely impossible in C++ or Java. 

If we write code for a class that uses this unbounded polymorphism, we want a way to inform the programmer as soon as possible if the class lacks a method that might be invoked polymorphically. 

Thanks, 
Ed

>>> Rick DeNatale <rick.denatale at gmail.com> 3/4/2011 1:29 PM >>>
IMHO, the right answer to questions like this is MU
http://en.wikipedia.org/wiki/Mu_(negative) ( http://en.wikipedia.org/wiki/Mu_(negative) )

This idea is cargo culting an idea from languages like C++ and Java.
If you want a statically typed language use one, Ruby is perfectly
fine without such stuff.

Of course I might be wrong <G>

On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
> We are going to use a module to act as an interface, as described in
> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby
> Is there any convention for naming an interface, e.g.,
>
> the interface for a contributor object should be called
> "contributor-interface"?
>
> Thanks,
> Ed
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>



--
Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Github: http://github.com/rubyredrick
Twitter: @RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale
_______________________________________________
raleigh-rb-members mailing list
raleigh-rb-members at rubyforge.org
http://rubyforge.org/mailman/listinfo/raleigh-rb-members

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/7d663323/attachment-0001.html>

From crnixon at gmail.com  Fri Mar  4 14:39:32 2011
From: crnixon at gmail.com (Clinton R. Nixon)
Date: Fri, 4 Mar 2011 14:39:32 -0500
Subject: [raleigh.rb] Interested in a talk on TorqueBox?
In-Reply-To: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>
References: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>
Message-ID: <AANLkTi=U9dQfiD-3os4Q5C-kkapft3S8DBwHfX89EFEZ@mail.gmail.com>

On Fri, Mar 4, 2011 at 11:04 AM, Tobias Crawley <tcrawley at gmail.com> wrote:
> I'm looking for opportunities to talk about TorqueBox (http://torquebox.org)
> - would you guys be interested?

I'd very much like to know more about TorqueBox. It's seemed
interesting to me, but a little overwhelming.

-- Clinton R. Nixon

From nathaniel at talbott.ws  Fri Mar  4 14:55:13 2011
From: nathaniel at talbott.ws (Nathaniel Talbott)
Date: Fri, 4 Mar 2011 14:55:13 -0500
Subject: [raleigh.rb] Interested in a talk on TorqueBox?
In-Reply-To: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>
References: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>
Message-ID: <AANLkTi=OXZMD_Bi=7SV_PbaTqm1zaLnDd4PYHYAnp3hN@mail.gmail.com>

On Fri, Mar 4, 2011 at 11:04 AM, Tobias Crawley <tcrawley at gmail.com> wrote:

> I'm looking for opportunities to talk about TorqueBox (http://torquebox.org)
> - would you guys be interested?
> If you haven't heard of TorqueBox - it's a ruby application server built on
> top of JBoss AS (using JRuby), and has integrated messaging, asynchronous
> processing, scheduled jobs, clustering, and more.
> I'm a member of the TorqueBox team. I live in Asheville, but would be happy
> to make the drive down.

Hey Toby, sounds intriguing, but: tell us a bit more why the average
Rubyist/Railsor should definitely be rocking some TorqueBox. I.e. how
might it change our behavior if we learn more about it?


-- 
Nathaniel Talbott
<:((><

From mriffe at gmail.com  Fri Mar  4 16:07:24 2011
From: mriffe at gmail.com (Mel Riffe)
Date: Fri, 4 Mar 2011 16:07:24 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <4D70F796.7D16.00A7.1@gw.ncsu.edu>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
Message-ID: <AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>

Ed,

I'm a little slow these days, please forgive me, but can you please explain:

We have a good application of unbounded polymorphism, where we have a
receiver that may be a member of classes that are not in the same
inheritance hierarchy. ?That's definitely impossible in C++ or Java.

I'm not familiar with the phrase 'unbounded polymorphism' or its mechanics.

Cheers,
Mel

On Fri, Mar 4, 2011 at 2:30 PM, Ed Gehringer <efg at ncsu.edu> wrote:
> I think you do have it wrong, Rick.
>
> We have a good application of unbounded polymorphism, where we have a
> receiver that may be a member of classes that are not in the same
> inheritance hierarchy. ?That's definitely impossible in C++ or Java.
>
> If we write code for a class that uses this unbounded polymorphism, we want
> a way to inform the programmer as soon as possible if the class lacks a
> method that might be invoked polymorphically.
>
> Thanks,
>
> Ed
>
>>>> Rick DeNatale <rick.denatale at gmail.com> 3/4/2011 1:29 PM >>>
>
> IMHO, the right answer to questions like this is MU
> http://en.wikipedia.org/wiki/Mu_(negative)
>
> This idea is cargo culting an idea from languages like C++ and Java.
> If you want a statically typed language use one, Ruby is perfectly
> fine without such stuff.
>
> Of course I might be wrong <G>
>
> On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
>> We are going to use a module to act as an interface, as described in
>> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby
>> Is there any convention for naming an interface, e.g.,
>>
>> the interface for a contributor object should be called
>> "contributor-interface"?
>>
>> Thanks,
>> Ed
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>
>
>
>
> --
> Rick DeNatale
>
> Blog: http://talklikeaduck.denhaven2.com/
> Github: http://github.com/rubyredrick
> Twitter: @RickDeNatale
> WWR: http://www.workingwithrails.com/person/9021-rick-denatale
> LinkedIn: http://www.linkedin.com/in/rickdenatale
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>

From tcrawley at gmail.com  Fri Mar  4 16:23:32 2011
From: tcrawley at gmail.com (Tobias Crawley)
Date: Fri, 4 Mar 2011 16:23:32 -0500
Subject: [raleigh.rb] Interested in a talk on TorqueBox?
In-Reply-To: <AANLkTi=OXZMD_Bi=7SV_PbaTqm1zaLnDd4PYHYAnp3hN@mail.gmail.com>
References: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>
	<AANLkTi=OXZMD_Bi=7SV_PbaTqm1zaLnDd4PYHYAnp3hN@mail.gmail.com>
Message-ID: <AANLkTikeLhiLQKz7-VGH6jkOdQWv+POWvhk2uNMfiQag@mail.gmail.com>

Sure thing:

TorqueBox is more than just a "webserver + ruby interpreter" - it's a real
ruby application server. So, out of the box, it provides (among other
things):

* asynchronous processing (using messaging) without having to setup and
manage DelayedJob/Resque/what-have-you
* scheduled jobs without dealing with external cron jobs
* clustering, with web session affinity, and automatic messaging
distribution

All of which are configured and managed from within your app codebase, and
share the lifecycle of your app. Therefore, if you undeploy or redeploy your
app, these components are automatically undeployed or redeployed along with
it, preventing them from getting out of sync, which can happen if they are
external services.

It uses JRuby, which provides speed, and proper threading. It is built on
top of JBoss AS 6, which gives is stable and performant.

Even though it is built on top of a Java application server, you can build
and deploy applications without touching a line of XML or Java.

It's under active development (we're rapidly approaching or 1.0RC1 release,
with 1.0Final targeted for May), and is in use in production by a handful of
'customers' at this point.

Is that enticing? It was to me, that's why I joined the TorqueBox team :)

Toby


On Fri, Mar 4, 2011 at 2:55 PM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:
>
>
> Hey Toby, sounds intriguing, but: tell us a bit more why the average
> Rubyist/Railsor should definitely be rocking some TorqueBox. I.e. how
> might it change our behavior if we learn more about it?
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/7b9eec48/attachment.html>

From korebantic at gmail.com  Fri Mar  4 17:03:22 2011
From: korebantic at gmail.com (korebantic)
Date: Fri, 4 Mar 2011 17:03:22 -0500
Subject: [raleigh.rb] Interested in a talk on TorqueBox?
In-Reply-To: <AANLkTikeLhiLQKz7-VGH6jkOdQWv+POWvhk2uNMfiQag@mail.gmail.com>
References: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>
	<AANLkTi=OXZMD_Bi=7SV_PbaTqm1zaLnDd4PYHYAnp3hN@mail.gmail.com>
	<AANLkTikeLhiLQKz7-VGH6jkOdQWv+POWvhk2uNMfiQag@mail.gmail.com>
Message-ID: <AANLkTinqPNQxOq4f0MePmtm+r3o0L+bqM7LXBt+M-MGf@mail.gmail.com>

I think it is (interesting). But I'm scared Steven may start the meeting
with some Yo dawgin' LOL

All BS aside +1 from me.

On Fri, Mar 4, 2011 at 4:23 PM, Tobias Crawley <tcrawley at gmail.com> wrote:

> Sure thing:
>
> TorqueBox is more than just a "webserver + ruby interpreter" - it's a real
> ruby application server. So, out of the box, it provides (among other
> things):
>
> * asynchronous processing (using messaging) without having to setup and
> manage DelayedJob/Resque/what-have-you
> * scheduled jobs without dealing with external cron jobs
> * clustering, with web session affinity, and automatic messaging
> distribution
>
> All of which are configured and managed from within your app codebase, and
> share the lifecycle of your app. Therefore, if you undeploy or redeploy your
> app, these components are automatically undeployed or redeployed along with
> it, preventing them from getting out of sync, which can happen if they are
> external services.
>
> It uses JRuby, which provides speed, and proper threading. It is built on
> top of JBoss AS 6, which gives is stable and performant.
>
> Even though it is built on top of a Java application server, you can build
> and deploy applications without touching a line of XML or Java.
>
> It's under active development (we're rapidly approaching or 1.0RC1 release,
> with 1.0Final targeted for May), and is in use in production by a handful of
> 'customers' at this point.
>
> Is that enticing? It was to me, that's why I joined the TorqueBox team :)
>
> Toby
>
>
> On Fri, Mar 4, 2011 at 2:55 PM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:
>>
>>
>> Hey Toby, sounds intriguing, but: tell us a bit more why the average
>> Rubyist/Railsor should definitely be rocking some TorqueBox. I.e. how
>> might it change our behavior if we learn more about it?
>>
>>
>>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/1ddf989f/attachment.html>

From motley.crue.fan at gmail.com  Fri Mar  4 18:59:26 2011
From: motley.crue.fan at gmail.com (Phillip Rhodes)
Date: Fri, 4 Mar 2011 18:59:26 -0500
Subject: [raleigh.rb] Interested in a talk on TorqueBox?
In-Reply-To: <AANLkTikeLhiLQKz7-VGH6jkOdQWv+POWvhk2uNMfiQag@mail.gmail.com>
References: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>
	<AANLkTi=OXZMD_Bi=7SV_PbaTqm1zaLnDd4PYHYAnp3hN@mail.gmail.com>
	<AANLkTikeLhiLQKz7-VGH6jkOdQWv+POWvhk2uNMfiQag@mail.gmail.com>
Message-ID: <AANLkTin_vbg6KG6E_ZCNd1sPGDoJQhRn2Xe7rC=x+2Zt@mail.gmail.com>

+1 here.

On Fri, Mar 4, 2011 at 4:23 PM, Tobias Crawley <tcrawley at gmail.com> wrote:

> Sure thing:
>
> TorqueBox is more than just a "webserver + ruby interpreter" - it's a real
> ruby application server. So, out of the box, it provides (among other
> things):
>
> * asynchronous processing (using messaging) without having to setup and
> manage DelayedJob/Resque/what-have-you
> * scheduled jobs without dealing with external cron jobs
> * clustering, with web session affinity, and automatic messaging
> distribution
>
> All of which are configured and managed from within your app codebase, and
> share the lifecycle of your app. Therefore, if you undeploy or redeploy your
> app, these components are automatically undeployed or redeployed along with
> it, preventing them from getting out of sync, which can happen if they are
> external services.
>
> It uses JRuby, which provides speed, and proper threading. It is built on
> top of JBoss AS 6, which gives is stable and performant.
>
> Even though it is built on top of a Java application server, you can build
> and deploy applications without touching a line of XML or Java.
>
> It's under active development (we're rapidly approaching or 1.0RC1 release,
> with 1.0Final targeted for May), and is in use in production by a handful of
> 'customers' at this point.
>
> Is that enticing? It was to me, that's why I joined the TorqueBox team :)
>
> Toby
>
>
> On Fri, Mar 4, 2011 at 2:55 PM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:
>>
>>
>> Hey Toby, sounds intriguing, but: tell us a bit more why the average
>> Rubyist/Railsor should definitely be rocking some TorqueBox. I.e. how
>> might it change our behavior if we learn more about it?
>>
>>
>>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/2178c5dd/attachment-0001.html>

From efg at ncsu.edu  Fri Mar  4 19:18:01 2011
From: efg at ncsu.edu (Ed Gehringer)
Date: Fri, 04 Mar 2011 19:18:01 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
Message-ID: <4D713AF0.7D16.00A7.1@gw.ncsu.edu>

I looked it up, and liked the first description I came across on the Web:  http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/237598 
 
Briefly, polymorphism in statically typed o-o languages depends on objects being in the same inheritance hierarchy.  If you have a method "perimeter" that is defined in class Shape, and if Circle, Square, and Trapezoid are subclasses of Shape, then objects of classes Circle, Square, Trapezoid, and Shape can invoke perimeter.  Other classes that are not subclasses of Shape cannot (for example, you can't call perimeter on a House object).
 
In Ruby and all other dynamic o-o languages that I know of, this is not true.  You could invoke the perimeter method on all the classes mentioned above, including House.  If House had a perimeter method, however implemented, the same statement that invoked perimeter on a Circle could also be used to invoke it on a House, providing that the receiver variable actually held a House at the time of the call.
 
This feature turned out to be very useful to us for a situation that would take a couple of paragraphs to explain.  I will spare you the details.
 
HTH,
Ed

>>> Mel Riffe <mriffe at gmail.com> 3/4/2011 4:07 PM >>>
Ed,

I'm a little slow these days, please forgive me, but can you please explain:

We have a good application of unbounded polymorphism, where we have a
receiver that may be a member of classes that are not in the same
inheritance hierarchy.  That's definitely impossible in C++ or Java.

I'm not familiar with the phrase 'unbounded polymorphism' or its mechanics.

Cheers,
Mel

On Fri, Mar 4, 2011 at 2:30 PM, Ed Gehringer <efg at ncsu.edu> wrote:
> I think you do have it wrong, Rick.
>
> We have a good application of unbounded polymorphism, where we have a
> receiver that may be a member of classes that are not in the same
> inheritance hierarchy.  That's definitely impossible in C++ or Java.
>
> If we write code for a class that uses this unbounded polymorphism, we want
> a way to inform the programmer as soon as possible if the class lacks a
> method that might be invoked polymorphically.
>
> Thanks,
>
> Ed
>
>>>> Rick DeNatale <rick.denatale at gmail.com> 3/4/2011 1:29 PM >>>
>
> IMHO, the right answer to questions like this is MU
> http://en.wikipedia.org/wiki/Mu_(negative)
>
> This idea is cargo culting an idea from languages like C++ and Java.
> If you want a statically typed language use one, Ruby is perfectly
> fine without such stuff.
>
> Of course I might be wrong <G>
>
> On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
>> We are going to use a module to act as an interface, as described in
>> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby 
>> Is there any convention for naming an interface, e.g.,
>>
>> the interface for a contributor object should be called
>> "contributor-interface"?
>>
>> Thanks,
>> Ed
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org 
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members 
>>
>
>
>
> --
> Rick DeNatale
>
> Blog: http://talklikeaduck.denhaven2.com/ 
> Github: http://github.com/rubyredrick 
> Twitter: @RickDeNatale
> WWR: http://www.workingwithrails.com/person/9021-rick-denatale 
> LinkedIn: http://www.linkedin.com/in/rickdenatale 
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org 
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members 
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org 
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members 
>
_______________________________________________
raleigh-rb-members mailing list
raleigh-rb-members at rubyforge.org 
http://rubyforge.org/mailman/listinfo/raleigh-rb-members
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/12eba500/attachment.html>

From steve at iannopollo.com  Fri Mar  4 21:27:22 2011
From: steve at iannopollo.com (Steve Iannopollo)
Date: Fri, 4 Mar 2011 21:27:22 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <4D713AF0.7D16.00A7.1@gw.ncsu.edu>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
Message-ID: <933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>

I'm pretty interested in your multi-paragraph explanation, because like Rick, I don't see any need to import a statically-typed language construct into a dynamic language. I checked out the chrisdavaz site to see what he was doing, and I'm wondering why nearly the same thing couldn't be accomplished in one shot (both interface and basic implementation methods) with something like the following:

module MyInterface
  def required_method
    raise NotImplementedError.new('ahhh!')
  end
end

class Thing
  include MyInterface
end

This doesn't blow up at load time, but it will only blow up when it needs to (which some may argue is the proper time to do so). And it would allow you to keep the include at the top of the class (which the chrisdavaz example doesn't do).

-Steve

On Mar 4, 2011, at 7:18 PM, Ed Gehringer wrote:

> I looked it up, and liked the first description I came across on the Web:  http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/237598
>  
> Briefly, polymorphism in statically typed o-o languages depends on objects being in the same inheritance hierarchy.  If you have a method "perimeter" that is defined in class Shape, and if Circle, Square, and Trapezoid are subclasses of Shape, then objects of classes Circle, Square, Trapezoid, and Shape can invoke perimeter.  Other classes that are not subclasses of Shape cannot (for example, you can't call perimeter on a House object).
>  
> In Ruby and all other dynamic o-o languages that I know of, this is not true.  You could invoke the perimeter method on all the classes mentioned above, including House.  If House had a perimeter method, however implemented, the same statement that invoked perimeter on a Circle could also be used to invoke it on a House, providing that the receiver variable actually held a House at the time of the call.
>  
> This feature turned out to be very useful to us for a situation that would take a couple of paragraphs to explain.  I will spare you the details.
>  
> HTH,
> Ed
> 
> >>> Mel Riffe <mriffe at gmail.com> 3/4/2011 4:07 PM >>>
> Ed,
> 
> I'm a little slow these days, please forgive me, but can you please explain:
> 
> We have a good application of unbounded polymorphism, where we have a
> receiver that may be a member of classes that are not in the same
> inheritance hierarchy.  That's definitely impossible in C++ or Java.
> 
> I'm not familiar with the phrase 'unbounded polymorphism' or its mechanics.
> 
> Cheers,
> Mel
> 
> On Fri, Mar 4, 2011 at 2:30 PM, Ed Gehringer <efg at ncsu.edu> wrote:
> > I think you do have it wrong, Rick.
> >
> > We have a good application of unbounded polymorphism, where we have a
> > receiver that may be a member of classes that are not in the same
> > inheritance hierarchy.  That's definitely impossible in C++ or Java.
> >
> > If we write code for a class that uses this unbounded polymorphism, we want
> > a way to inform the programmer as soon as possible if the class lacks a
> > method that might be invoked polymorphically.
> >
> > Thanks,
> >
> > Ed
> >
> >>>> Rick DeNatale <rick.denatale at gmail.com> 3/4/2011 1:29 PM >>>
> >
> > IMHO, the right answer to questions like this is MU
> > http://en.wikipedia.org/wiki/Mu_(negative)
> >
> > This idea is cargo culting an idea from languages like C++ and Java.
> > If you want a statically typed language use one, Ruby is perfectly
> > fine without such stuff.
> >
> > Of course I might be wrong <G>
> >
> > On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
> >> We are going to use a module to act as an interface, as described in
> >> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby
> >> Is there any convention for naming an interface, e.g.,
> >>
> >> the interface for a contributor object should be called
> >> "contributor-interface"?
> >>
> >> Thanks,
> >> Ed
> >> _______________________________________________
> >> raleigh-rb-members mailing list
> >> raleigh-rb-members at rubyforge.org
> >> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
> >>
> >
> >
> >
> > --
> > Rick DeNatale
> >
> > Blog: http://talklikeaduck.denhaven2.com/
> > Github: http://github.com/rubyredrick
> > Twitter: @RickDeNatale
> > WWR: http://www.workingwithrails.com/person/9021-rick-denatale
> > LinkedIn: http://www.linkedin.com/in/rickdenatale
> > _______________________________________________
> > raleigh-rb-members mailing list
> > raleigh-rb-members at rubyforge.org
> > http://rubyforge.org/mailman/listinfo/raleigh-rb-members
> >
> > _______________________________________________
> > raleigh-rb-members mailing list
> > raleigh-rb-members at rubyforge.org
> > http://rubyforge.org/mailman/listinfo/raleigh-rb-members
> >
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/28b6335b/attachment.html>

From efg at ncsu.edu  Fri Mar  4 22:18:41 2011
From: efg at ncsu.edu (Ed Gehringer)
Date: Fri, 04 Mar 2011 22:18:41 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
Message-ID: <4D716549.7D16.00A7.1@gw.ncsu.edu>

Hey Steve,
 
I'll get to this tomorrow ... it will take a little time.
 
As to the module you propose, that's a good solution if there's one required method.  If there are 10, it gets a little repetitive, so I like the Chris Davaz approach better.
 
-Ed

>>> Steve Iannopollo <steve at iannopollo.com> 3/4/2011 9:27 PM >>>
I'm pretty interested in your multi-paragraph explanation, because like Rick, I don't see any need to import a statically-typed language construct into a dynamic language. I checked out the chrisdavaz site to see what he was doing, and I'm wondering why nearly the same thing couldn't be accomplished in one shot (both interface and basic implementation methods) with something like the following:

module MyInterface
  def required_method
    raise NotImplementedError.new('ahhh!')
  end
end

class Thing
  include MyInterface
end

This doesn't blow up at load time, but it will only blow up when it needs to (which some may argue is the proper time to do so). And it would allow you to keep the include at the top of the class (which the chrisdavaz example doesn't do).

-Steve

On Mar 4, 2011, at 7:18 PM, Ed Gehringer wrote:



I looked it up, and liked the first description I came across on the Web:  http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/237598 
 
Briefly, polymorphism in statically typed o-o languages depends on objects being in the same inheritance hierarchy.  If you have a method "perimeter" that is defined in class Shape, and if Circle, Square, and Trapezoid are subclasses of Shape, then objects of classes Circle, Square, Trapezoid, and Shape can invoke perimeter.  Other classes that are not subclasses of Shape cannot (for example, you can't call perimeter on a House object).
 
In Ruby and all other dynamic o-o languages that I know of, this is not true.  You could invoke the perimeter method on all the classes mentioned above, including House.  If House had a perimeter method, however implemented, the same statement that invoked perimeter on a Circle could also be used to invoke it on a House, providing that the receiver variable actually held a House at the time of the call.
 
This feature turned out to be very useful to us for a situation that would take a couple of paragraphs to explain.  I will spare you the details.
 
HTH,
Ed

>>> Mel Riffe <mriffe at gmail.com> 3/4/2011 4:07 PM >>>
Ed,

I'm a little slow these days, please forgive me, but can you please explain:

We have a good application of unbounded polymorphism, where we have a
receiver that may be a member of classes that are not in the same
inheritance hierarchy.  That's definitely impossible in C++ or Java.

I'm not familiar with the phrase 'unbounded polymorphism' or its mechanics.

Cheers,
Mel

On Fri, Mar 4, 2011 at 2:30 PM, Ed Gehringer <efg at ncsu.edu> wrote:
> I think you do have it wrong, Rick.
>
> We have a good application of unbounded polymorphism, where we have a
> receiver that may be a member of classes that are not in the same
> inheritance hierarchy.  That's definitely impossible in C++ or Java.
>
> If we write code for a class that uses this unbounded polymorphism, we want
> a way to inform the programmer as soon as possible if the class lacks a
> method that might be invoked polymorphically.
>
> Thanks,
>
> Ed
>
>>>> Rick DeNatale <rick.denatale at gmail.com> 3/4/2011 1:29 PM >>>
>
> IMHO, the right answer to questions like this is MU
> http://en.wikipedia.org/wiki/Mu_(negative)
>
> This idea is cargo culting an idea from languages like C++ and Java.
> If you want a statically typed language use one, Ruby is perfectly
> fine without such stuff.
>
> Of course I might be wrong <G>
>
> On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
>> We are going to use a module to act as an interface, as described in
>> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby 
>> Is there any convention for naming an interface, e.g.,
>>
>> the interface for a contributor object should be called
>> "contributor-interface"?
>>
>> Thanks,
>> Ed
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org 
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members 
>>
>
>
>
> --
> Rick DeNatale
>
> Blog: http://talklikeaduck.denhaven2.com/ 
> Github: http://github.com/rubyredrick 
> Twitter: @RickDeNatale
> WWR: http://www.workingwithrails.com/person/9021-rick-denatale 
> LinkedIn: http://www.linkedin.com/in/rickdenatale 
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org 
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members 
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org 
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members 
>
_______________________________________________
raleigh-rb-members mailing list
raleigh-rb-members at rubyforge.org 
http://rubyforge.org/mailman/listinfo/raleigh-rb-members 
_______________________________________________
raleigh-rb-members mailing list
raleigh-rb-members at rubyforge.org 
http://rubyforge.org/mailman/listinfo/raleigh-rb-members 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110304/7ffecd02/attachment-0001.html>

From steve at iannopollo.com  Fri Mar  4 23:24:33 2011
From: steve at iannopollo.com (Steve Iannopollo)
Date: Fri, 4 Mar 2011 23:24:33 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <4D716549.7D16.00A7.1@gw.ncsu.edu>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
	<4D716549.7D16.00A7.1@gw.ncsu.edu>
Message-ID: <1EC59F74-C645-4A0F-8FEF-CC93156A7941@iannopollo.com>

The concept could still be applied though:

module MyInterface
  %w(method1 method2 method3).each do |method_name|
    define_method method_name do
      raise NotImplementedError.new("You need to implement ##{method_name}")
    end
  end
end

There always seems to be more than one way to skin the proverbial cat in ruby :-)

-Steve

On Mar 4, 2011, at 10:18 PM, Ed Gehringer wrote:

> Hey Steve,
>  
> I'll get to this tomorrow ... it will take a little time.
>  
> As to the module you propose, that's a good solution if there's one required method.  If there are 10, it gets a little repetitive, so I like the Chris Davaz approach better.
>  
> -Ed
> 
> >>> Steve Iannopollo <steve at iannopollo.com> 3/4/2011 9:27 PM >>>
> I'm pretty interested in your multi-paragraph explanation, because like Rick, I don't see any need to import a statically-typed language construct into a dynamic language. I checked out the chrisdavaz site to see what he was doing, and I'm wondering why nearly the same thing couldn't be accomplished in one shot (both interface and basic implementation methods) with something like the following:
> 
> module MyInterface
>   def required_method
>     raise NotImplementedError.new('ahhh!')
>   end
> end
> 
> class Thing
>   include MyInterface
> end
> 
> This doesn't blow up at load time, but it will only blow up when it needs to (which some may argue is the proper time to do so). And it would allow you to keep the include at the top of the class (which the chrisdavaz example doesn't do).
> 
> -Steve
> 
> On Mar 4, 2011, at 7:18 PM, Ed Gehringer wrote:
> 
>> I looked it up, and liked the first description I came across on the Web:  http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/237598
>>  
>> Briefly, polymorphism in statically typed o-o languages depends on objects being in the same inheritance hierarchy.  If you have a method "perimeter" that is defined in class Shape, and if Circle, Square, and Trapezoid are subclasses of Shape, then objects of classes Circle, Square, Trapezoid, and Shape can invoke perimeter.  Other classes that are not subclasses of Shape cannot (for example, you can't call perimeter on a House object).
>>  
>> In Ruby and all other dynamic o-o languages that I know of, this is not true.  You could invoke the perimeter method on all the classes mentioned above, including House.  If House had a perimeter method, however implemented, the same statement that invoked perimeter on a Circle could also be used to invoke it on a House, providing that the receiver variable actually held a House at the time of the call.
>>  
>> This feature turned out to be very useful to us for a situation that would take a couple of paragraphs to explain.  I will spare you the details.
>>  
>> HTH,
>> Ed
>> 
>> >>> Mel Riffe <mriffe at gmail.com> 3/4/2011 4:07 PM >>>
>> Ed,
>> 
>> I'm a little slow these days, please forgive me, but can you please explain:
>> 
>> We have a good application of unbounded polymorphism, where we have a
>> receiver that may be a member of classes that are not in the same
>> inheritance hierarchy.  That's definitely impossible in C++ or Java.
>> 
>> I'm not familiar with the phrase 'unbounded polymorphism' or its mechanics.
>> 
>> Cheers,
>> Mel
>> 
>> On Fri, Mar 4, 2011 at 2:30 PM, Ed Gehringer <efg at ncsu.edu> wrote:
>> > I think you do have it wrong, Rick.
>> >
>> > We have a good application of unbounded polymorphism, where we have a
>> > receiver that may be a member of classes that are not in the same
>> > inheritance hierarchy.  That's definitely impossible in C++ or Java.
>> >
>> > If we write code for a class that uses this unbounded polymorphism, we want
>> > a way to inform the programmer as soon as possible if the class lacks a
>> > method that might be invoked polymorphically.
>> >
>> > Thanks,
>> >
>> > Ed
>> >
>> >>>> Rick DeNatale <rick.denatale at gmail.com> 3/4/2011 1:29 PM >>>
>> >
>> > IMHO, the right answer to questions like this is MU
>> > http://en.wikipedia.org/wiki/Mu_(negative)
>> >
>> > This idea is cargo culting an idea from languages like C++ and Java.
>> > If you want a statically typed language use one, Ruby is perfectly
>> > fine without such stuff.
>> >
>> > Of course I might be wrong <G>
>> >
>> > On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
>> >> We are going to use a module to act as an interface, as described in
>> >> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby
>> >> Is there any convention for naming an interface, e.g.,
>> >>
>> >> the interface for a contributor object should be called
>> >> "contributor-interface"?
>> >>
>> >> Thanks,
>> >> Ed
>> >> _______________________________________________
>> >> raleigh-rb-members mailing list
>> >> raleigh-rb-members at rubyforge.org
>> >> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>> >>
>> >
>> >
>> >
>> > --
>> > Rick DeNatale
>> >
>> > Blog: http://talklikeaduck.denhaven2.com/
>> > Github: http://github.com/rubyredrick
>> > Twitter: @RickDeNatale
>> > WWR: http://www.workingwithrails.com/person/9021-rick-denatale
>> > LinkedIn: http://www.linkedin.com/in/rickdenatale
>> > _______________________________________________
>> > raleigh-rb-members mailing list
>> > raleigh-rb-members at rubyforge.org
>> > http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>> >
>> > _______________________________________________
>> > raleigh-rb-members mailing list
>> > raleigh-rb-members at rubyforge.org
>> > http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>> >
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
> 
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members


From mshiltonj at gmail.com  Sat Mar  5 03:17:58 2011
From: mshiltonj at gmail.com (Steven Hilton)
Date: Sat, 5 Mar 2011 03:17:58 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
Message-ID: <AANLkTimu65f6YOE5-Mo2s96QxT7CE-wzSsh5tPJvEjfP@mail.gmail.com>

Wouldn't it blow up anyway, with a NoMethodError?

On Fri, Mar 4, 2011 at 9:27 PM, Steve Iannopollo <steve at iannopollo.com> wrote:
> I'm pretty interested in your multi-paragraph explanation, because like
> Rick, I don't see any need to import a statically-typed language construct
> into a dynamic language. I checked out the chrisdavaz site to see what he
> was doing, and I'm wondering why nearly the same thing couldn't be
> accomplished in one shot (both interface and basic implementation methods)
> with something like the following:
> module MyInterface
> ??def required_method
> ?? ?raise NotImplementedError.new('ahhh!')
> ??end
> end
> class Thing
> ??include MyInterface
> end
> This doesn't blow up at load time, but it will only blow up when it needs to
> (which some may argue is the proper time to do so). And it would allow you
> to keep the include at the top of the class (which the?chrisdavaz example
> doesn't do).
>
> -Steve
> On Mar 4, 2011, at 7:18 PM, Ed Gehringer wrote:
>
> I looked it up, and liked the first description I came across on the Web:
> http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/237598
>
> Briefly, polymorphism in statically typed o-o languages depends on objects
> being in the same inheritance hierarchy.? If you have a method "perimeter"
> that is defined in class Shape, and if Circle, Square, and Trapezoid are
> subclasses of Shape, then objects of classes Circle, Square, Trapezoid, and
> Shape can invoke perimeter.? Other classes that are not subclasses of Shape
> cannot (for example, you can't call perimeter on a House object).
>
> In Ruby and all other dynamic o-o languages that I know of, this is not
> true.? You could invoke the perimeter method on all the classes mentioned
> above, including House.? If House had a perimeter method, however
> implemented, the same statement that invoked perimeter on a Circle could
> also be used to invoke it on a House, providing that the receiver variable
> actually held a House at the time of the call.
>
> This feature turned out to be very useful to us for a situation that would
> take a couple of paragraphs to explain.? I will spare you the details.
>
> HTH,
> Ed
>
>>>> Mel Riffe <mriffe at gmail.com> 3/4/2011 4:07 PM >>>
> Ed,
>
> I'm a little slow these days, please forgive me, but can you please explain:
>
> We have a good application of unbounded polymorphism, where we have a
> receiver that may be a member of classes that are not in the same
> inheritance hierarchy.? That's definitely impossible in C++ or Java.
>
> I'm not familiar with the phrase 'unbounded polymorphism' or its mechanics.
>
> Cheers,
> Mel
>
> On Fri, Mar 4, 2011 at 2:30 PM, Ed Gehringer <efg at ncsu.edu> wrote:
>> I think you do have it wrong, Rick.
>>
>> We have a good application of unbounded polymorphism, where we have a
>> receiver that may be a member of classes that are not in the same
>> inheritance hierarchy.? That's definitely impossible in C++ or Java.
>>
>> If we write code for a class that uses this unbounded polymorphism, we
>> want
>> a way to inform the programmer as soon as possible if the class lacks a
>> method that might be invoked polymorphically.
>>
>> Thanks,
>>
>> Ed
>>
>>>>> Rick DeNatale <rick.denatale at gmail.com> 3/4/2011 1:29 PM >>>
>>
>> IMHO, the right answer to questions like this is MU
>> http://en.wikipedia.org/wiki/Mu_(negative)
>>
>> This idea is cargo culting an idea from languages like C++ and Java.
>> If you want a statically typed language use one, Ruby is perfectly
>> fine without such stuff.
>>
>> Of course I might be wrong <G>
>>
>> On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
>>> We are going to use a module to act as an interface, as described in
>>> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby
>>> Is there any convention for naming an interface, e.g.,
>>>
>>> the interface for a contributor object should be called
>>> "contributor-interface"?
>>>
>>> Thanks,
>>> Ed
>>> _______________________________________________
>>> raleigh-rb-members mailing list
>>> raleigh-rb-members at rubyforge.org
>>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>>
>>
>>
>>
>> --
>> Rick DeNatale
>>
>> Blog: http://talklikeaduck.denhaven2.com/
>> Github: http://github.com/rubyredrick
>> Twitter: @RickDeNatale
>> WWR: http://www.workingwithrails.com/person/9021-rick-denatale
>> LinkedIn: http://www.linkedin.com/in/rickdenatale
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>

From jeff at jumpstartlab.com  Sat Mar  5 11:48:54 2011
From: jeff at jumpstartlab.com (Jeff Casimir)
Date: Sat, 5 Mar 2011 11:48:54 -0500
Subject: [raleigh.rb] Early Bird Ends TODAY for Ruby and Rails Jumpstart
	Classes
Message-ID: <AANLkTinVSUmnuE2RrSAs4qWNvRZwAMvKOkp+-eTMmTS9@mail.gmail.com>

RaleighRB,

It was fun hanging out in your town this week and meeting several of
you between my Ruby Primer and your Hack Night.  I look forward to
coming back down in a few weeks and hopefully having more time to see
the cities.

I wanted to remind you that today is the last day for Early Bird
tickets for my two classes.

The full two-day Ruby class will take place 3/26-3/27.  We'll spend
one day on a data
manipulation project, cleaning up registration data from a conference
and using it to access government representative information from the
Sunlight Foundation.  In the second day we'll build a command-line
Twitter client to play around with accessing an external service and
working with deeply nested data.  This class is built for
non-programmers, so if you don't have any experience or it's been a
long time, this is for you:
http://jumpstartlab.com/trainings/14-ruby-raleigh

Then the following two days (3/28-3/29) I'll be teaching Intro to
Rails.  This course is really for existing programmers.  If you're
brand new to programming and take the Intro to Ruby class, you can
just barely hang with Intro Rails, but it'll be tough!  We're going to
build a Rails application from the ground up, building your
understanding of each component in the Rails architecture and how they
fit together.  Tickets are available here:
http://jumpstartlab.com/trainings/15-rails-raleigh

For both classes the early bird registration ends TODAY!  You're
welcome to use and share the discount code "raleighrb" which will take
an additional 10% off the ticket price.

Let me know if you have any questions, I look forward to seeing you in
a few weeks!

- Jeff

---
Jeff Casimir
Jumpstart Lab by Casimir Creative, LLC
http://jumpstartlab.com
@jumpstartlab on twitter

From steve at iannopollo.com  Sat Mar  5 13:07:16 2011
From: steve at iannopollo.com (Steve Iannopollo)
Date: Sat, 5 Mar 2011 13:07:16 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <AANLkTimu65f6YOE5-Mo2s96QxT7CE-wzSsh5tPJvEjfP@mail.gmail.com>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
	<AANLkTimu65f6YOE5-Mo2s96QxT7CE-wzSsh5tPJvEjfP@mail.gmail.com>
Message-ID: <67FB99FA-6B87-4890-8CA6-23967ECBEEE5@iannopollo.com>

> Wouldn't it blow up anyway, with a NoMethodError?

So with the following code:

>> module MyInterface
>>   def required_method
>>     raise NotImplementedError.new('ahhh!')
>>   end
>> end
>> class Thing
>>   include MyInterface
>> end

You can call

thing = Thing.new
thing.required_method

and that will raise a NotImplementedError. With the include of MyInterface in the Thing class, you already get #required_method defined for you, but it raises the error letting you know something is wrong. That should force you to redefine #required_method on the Thing class.

Sorry dawg, I don't like the javaz in my ruby :-)

-Steve


>> This doesn't blow up at load time, but it will only blow up when it needs to
>> (which some may argue is the proper time to do so). And it would allow you
>> to keep the include at the top of the class (which the chrisdavaz example
>> doesn't do).
>> 
>> -Steve
>> On Mar 4, 2011, at 7:18 PM, Ed Gehringer wrote:
>> 
>> I looked it up, and liked the first description I came across on the Web:
>> http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/237598
>> 
>> Briefly, polymorphism in statically typed o-o languages depends on objects
>> being in the same inheritance hierarchy.  If you have a method "perimeter"
>> that is defined in class Shape, and if Circle, Square, and Trapezoid are
>> subclasses of Shape, then objects of classes Circle, Square, Trapezoid, and
>> Shape can invoke perimeter.  Other classes that are not subclasses of Shape
>> cannot (for example, you can't call perimeter on a House object).
>> 
>> In Ruby and all other dynamic o-o languages that I know of, this is not
>> true.  You could invoke the perimeter method on all the classes mentioned
>> above, including House.  If House had a perimeter method, however
>> implemented, the same statement that invoked perimeter on a Circle could
>> also be used to invoke it on a House, providing that the receiver variable
>> actually held a House at the time of the call.
>> 
>> This feature turned out to be very useful to us for a situation that would
>> take a couple of paragraphs to explain.  I will spare you the details.
>> 
>> HTH,
>> Ed
>> 
>>>>> Mel Riffe <mriffe at gmail.com> 3/4/2011 4:07 PM >>>
>> Ed,
>> 
>> I'm a little slow these days, please forgive me, but can you please explain:
>> 
>> We have a good application of unbounded polymorphism, where we have a
>> receiver that may be a member of classes that are not in the same
>> inheritance hierarchy.  That's definitely impossible in C++ or Java.
>> 
>> I'm not familiar with the phrase 'unbounded polymorphism' or its mechanics.
>> 
>> Cheers,
>> Mel
>> 
>> On Fri, Mar 4, 2011 at 2:30 PM, Ed Gehringer <efg at ncsu.edu> wrote:
>>> I think you do have it wrong, Rick.
>>> 
>>> We have a good application of unbounded polymorphism, where we have a
>>> receiver that may be a member of classes that are not in the same
>>> inheritance hierarchy.  That's definitely impossible in C++ or Java.
>>> 
>>> If we write code for a class that uses this unbounded polymorphism, we
>>> want
>>> a way to inform the programmer as soon as possible if the class lacks a
>>> method that might be invoked polymorphically.
>>> 
>>> Thanks,
>>> 
>>> Ed
>>> 
>>>>>> Rick DeNatale <rick.denatale at gmail.com> 3/4/2011 1:29 PM >>>
>>> 
>>> IMHO, the right answer to questions like this is MU
>>> http://en.wikipedia.org/wiki/Mu_(negative)
>>> 
>>> This idea is cargo culting an idea from languages like C++ and Java.
>>> If you want a statically typed language use one, Ruby is perfectly
>>> fine without such stuff.
>>> 
>>> Of course I might be wrong <G>
>>> 
>>> On Fri, Mar 4, 2011 at 11:47 AM, Ed Gehringer <efg at ncsu.edu> wrote:
>>>> We are going to use a module to act as an interface, as described in
>>>> http://www.chrisdavaz.com/blog/how-to-enforce-interfaces-in-ruby
>>>> Is there any convention for naming an interface, e.g.,
>>>> 
>>>> the interface for a contributor object should be called
>>>> "contributor-interface"?
>>>> 
>>>> Thanks,
>>>> Ed
>>>> _______________________________________________
>>>> raleigh-rb-members mailing list
>>>> raleigh-rb-members at rubyforge.org
>>>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>>> 
>>> 
>>> 
>>> 
>>> --
>>> Rick DeNatale
>>> 
>>> Blog: http://talklikeaduck.denhaven2.com/
>>> Github: http://github.com/rubyredrick
>>> Twitter: @RickDeNatale
>>> WWR: http://www.workingwithrails.com/person/9021-rick-denatale
>>> LinkedIn: http://www.linkedin.com/in/rickdenatale
>>> _______________________________________________
>>> raleigh-rb-members mailing list
>>> raleigh-rb-members at rubyforge.org
>>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>> 
>>> _______________________________________________
>>> raleigh-rb-members mailing list
>>> raleigh-rb-members at rubyforge.org
>>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>> 
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>> 
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>> 


From nathaniel at talbott.ws  Sat Mar  5 15:58:52 2011
From: nathaniel at talbott.ws (Nathaniel Talbott)
Date: Sat, 5 Mar 2011 15:58:52 -0500
Subject: [raleigh.rb] Interested in a talk on TorqueBox?
In-Reply-To: <AANLkTikeLhiLQKz7-VGH6jkOdQWv+POWvhk2uNMfiQag@mail.gmail.com>
References: <AANLkTi=u4joj=LsSvTS3DyAhK5UEVX0w0VKkKKdv-o8T@mail.gmail.com>
	<AANLkTi=OXZMD_Bi=7SV_PbaTqm1zaLnDd4PYHYAnp3hN@mail.gmail.com>
	<AANLkTikeLhiLQKz7-VGH6jkOdQWv+POWvhk2uNMfiQag@mail.gmail.com>
Message-ID: <AANLkTi=3hQQDcnM04O1WYA3bQyGYptjbnp3m2ODewRgw@mail.gmail.com>

On Fri, Mar 4, 2011 at 4:23 PM, Tobias Crawley <tcrawley at gmail.com> wrote:

> Is that enticing? It was to me, that's why I joined the TorqueBox team :)

Nicely done :-)

I'll email you directly so we can work out details and get you scheduled.


-- 
Nathaniel Talbott
<:((><

From efg at ncsu.edu  Sat Mar  5 23:54:22 2011
From: efg at ncsu.edu (Ed Gehringer)
Date: Sat, 05 Mar 2011 23:54:22 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
Message-ID: <4D72CD39.7D16.00A7.1@gw.ncsu.edu>

>>> Steve Iannopollo <steve at iannopollo.com> 3/4/2011 9:27 PM >>>
I'm pretty interested in your multi-paragraph explanation, because like Rick, I don't see any need to import a statically-typed language construct into a dynamic language. 
 
<<<
 
OK, here goes.
 
The application is our Expertiza collaborative learning system.  The code needs to assign reviewers for work that is produced either by an individual author or a team.
 
If an individual author is being reviewed, the code finds information about the author (user-ID, topic that the author is writing on, links to submitted work, etc.) in the participants table.  If a team is being reviewed, relevant information (team name, etc.) is found in the teams table.  Obviously, author should not be a subclass of team, or vice versa.  But both author and team need to respond to the same messages.  A further complication is the fact that a participant can perform some functions (e.g., selecting a topic) as an individual, and then form a team, which "inherits" the topic that the participant has chosen.
 
Thus, both Participant and Team need to respond to a common set of messages.  (Both Participant and Team also need to respond to their own peculiar messages, which are not in this set.)  The standard way to do this in Java, C#, etc., would be to have Participant and Team implement the same interface.  In Ruby, they simply need to implement the common set of methods.  Nonetheless, it will be useful to group these methods somewhere in the code, for the sake of readability and extensibility.
 
-Ed
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110305/75164f71/attachment.html>

From rick.denatale at gmail.com  Sun Mar  6 07:29:11 2011
From: rick.denatale at gmail.com (Rick DeNatale)
Date: Sun, 6 Mar 2011 07:29:11 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <4D72CD39.7D16.00A7.1@gw.ncsu.edu>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
	<4D72CD39.7D16.00A7.1@gw.ncsu.edu>
Message-ID: <AANLkTinJagOni3=jNf9gsuJ7yuB7AWLZpg8oUnK+Nv2-@mail.gmail.com>

On Sat, Mar 5, 2011 at 11:54 PM, Ed Gehringer <efg at ncsu.edu> wrote:
>>>> Steve Iannopollo <steve at iannopollo.com> 3/4/2011 9:27 PM >>>
> I'm pretty interested in your multi-paragraph explanation, because like
> Rick, I don't see any need to import a statically-typed language construct
> into a dynamic language.
>
> <<<
>
> OK, here goes.
>
> The application is our Expertiza collaborative learning system.? The code
> needs to assign reviewers for work that is produced either by an individual
> author or a team.
>
> If an individual author is being reviewed, the code finds information about
> the author (user-ID, topic that the author is writing on, links to submitted
> work, etc.) in the participants table.? If a team is being reviewed,
> relevant information (team name, etc.) is found in the teams table.
> Obviously, author should not be a subclass of team, or vice versa.? But both
> author and team need to respond to the same messages.? A further
> complication is the fact that a participant can perform some functions
> (e.g., selecting a topic) as an individual, and then form a team, which
> "inherits" the topic that the participant has chosen.
>
> Thus, both Participant and Team need to respond to a common set of
> messages.? (Both Participant and Team also need to respond to their own
> peculiar messages, which are not in this set.)? The standard way to do this
> in Java, C#, etc., would be to have Participant and Team implement the same
> interface.? In Ruby, they simply need to implement the common set of
> methods.? Nonetheless, it will be useful to group these methods somewhere in
> the code, for the sake of readability and extensibility.

It occur to me that this would be simpler if you just modeled it as a
work always being authored by a team, where a team has ONE or more
participants, rather than treating single or multiple authorships as
different cases.

In any event, I'd user specs/tests rather than trying to build some
kind of static typing kludge on top of Ruby.  Use modules if there's
reusable IMPLEMENTATION of methods, if you really need multiple
inheritance, otherwise just give each class an implementation of the
needed methods.

-- 
Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Github: http://github.com/rubyredrick
Twitter: @RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale

From skmetz at gmail.com  Sun Mar  6 07:56:09 2011
From: skmetz at gmail.com (Sandi Metz)
Date: Sun, 6 Mar 2011 07:56:09 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <4D72CD39.7D16.00A7.1@gw.ncsu.edu>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
	<4D72CD39.7D16.00A7.1@gw.ncsu.edu>
Message-ID: <AANLkTik5BRiTPjWxe1-A+9By5wESFuseOjdP-WeER9A9@mail.gmail.com>

You can use the tests you've written for your modules to prove that the
classes that include those modules honor the interface.
http://relishapp.com/rspec/rspec-core/v/2-0/dir/example-groups/shared-example-group

In your case you might have a contributor_spec, which defined some

shared_examples_for "contributor"

and then each class that included the contributor module would assert
something like

describe Participant
   it_should_behave_like "contributor"
   ...


The tests ought to prove that everything is ok AND provide all of the
documentation you need.
If they don't, improve the tests.
Win-win.

Of course, this assumes that you're using rspec  :-) but surely something
similar is possible in other test frameworks.

-Sandi

On Sat, Mar 5, 2011 at 11:54 PM, Ed Gehringer <efg at ncsu.edu> wrote:

>  >>> Steve Iannopollo <steve at iannopollo.com> 3/4/2011 9:27 PM >>>
> I'm pretty interested in your multi-paragraph explanation, because like
> Rick, I don't see any need to import a statically-typed language construct
> into a dynamic language.
>
> <<<
>
> OK, here goes.
>
> The application is our Expertiza collaborative learning system.  The code
> needs to assign reviewers for work that is produced either by an individual
> author or a team.
>
> If an individual author is being reviewed, the code finds information about
> the author (user-ID, topic that the author is writing on, links to submitted
> work, etc.) in the *participants* table.  If a team is being reviewed,
> relevant information (team name, etc.) is found in the *teams* table.
> Obviously, author should not be a subclass of team, or vice versa.  But both
> author and team need to respond to the same messages.  A further
> complication is the fact that a participant can perform some functions
> (e.g., selecting a topic) as an individual, and then form a team, which
> "inherits" the topic that the participant has chosen.
>
> Thus, both Participant and Team need to respond to a common set of
> messages.  (Both Participant and Team also need to respond to their own
> peculiar messages, which are not in this set.)  The standard way to do this
> in Java, C#, etc., would be to have Participant and Team implement the same
> interface.  In Ruby, they simply need to implement the common set of
> methods.  Nonetheless, it will be useful to group these methods somewhere in
> the code, for the sake of readability and extensibility.
>
> -Ed
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110306/585a3caf/attachment.html>

From trent.albright at gmail.com  Sun Mar  6 09:54:47 2011
From: trent.albright at gmail.com (Trent Albright)
Date: Sun, 6 Mar 2011 09:54:47 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <AANLkTik5BRiTPjWxe1-A+9By5wESFuseOjdP-WeER9A9@mail.gmail.com>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
	<4D72CD39.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTik5BRiTPjWxe1-A+9By5wESFuseOjdP-WeER9A9@mail.gmail.com>
Message-ID: <AANLkTikbdS5n1seNdA2r6hxGNQ5WihjMP=12eCpm9nbP@mail.gmail.com>

I've written plenty of production code in Java. I prefer by far duck typing
in Ruby than having to declare and fulfill interface contracts:

*In duck typing, one is concerned with just those aspects of an object that
are used, rather than with the type of the object itself.*


It feels like interfaces in Java are about compile time checking and
something that the VM runtime cares about. From a developers perspective,
seeing "implements MyInterface" in Java lets me know what kind of behavior
the class has or that I should provide. It's an arbitrary contract that
needs to be fulfilled. But I think all we really care about is whether a
given object can receive a given message. We don't need interfaces in Ruby
to accomplish that.

In general, in Ruby when we want to make sure objects behave as they are
supposed to we write specs/tests. In cases where you need to check at
runtime we use "Object#respond_to?"

I agree completely with Rick and Sandi.

You can use the tests you've written for your modules to prove that the
> classes that include those modules honor the interface.
>
> http://relishapp.com/rspec/rspec-core/v/2-0/dir/example-groups/shared-example-group
>
> In your case you might have a contributor_spec, which defined some
>
> shared_examples_for "contributor"
>
> and then each class that included the contributor module would assert
> something like
>
> describe Participant
>    it_should_behave_like "contributor"
>    ...
>
>
> The tests ought to prove that everything is ok AND provide all of the
> documentation you need.
> If they don't, improve the tests.
> Win-win.
>
> Of course, this assumes that you're using rspec  :-) but surely something
> similar is possible in other test frameworks.
>
> -Sandi
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110306/42daa95e/attachment.html>

From seancribbs at gmail.com  Sun Mar  6 12:52:12 2011
From: seancribbs at gmail.com (Sean Cribbs)
Date: Sun, 6 Mar 2011 12:52:12 -0500
Subject: [raleigh.rb] Convention for naming modules used as interfaces
In-Reply-To: <AANLkTikbdS5n1seNdA2r6hxGNQ5WihjMP=12eCpm9nbP@mail.gmail.com>
References: <4D70D171.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTi=5205dWcMwk0yJAyG0OSh1TrWYMzmAQX2vnP2Q@mail.gmail.com>
	<4D70F796.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTim-E6g6JmQq6PZkW8WS6pimYVet-h+zPW-1OUmr@mail.gmail.com>
	<4D713AF0.7D16.00A7.1@gw.ncsu.edu>
	<933453D2-FB0B-4A96-B3E2-C7C636C2B625@iannopollo.com>
	<4D72CD39.7D16.00A7.1@gw.ncsu.edu>
	<AANLkTik5BRiTPjWxe1-A+9By5wESFuseOjdP-WeER9A9@mail.gmail.com>
	<AANLkTikbdS5n1seNdA2r6hxGNQ5WihjMP=12eCpm9nbP@mail.gmail.com>
Message-ID: <AANLkTikf6-_0R2r2AjiVU9G+Y3Qe9MdQoLFjWOFwOLca@mail.gmail.com>

Interestingly, I've been more interested in having "contracts" of the
Eiffel sort lately.  We have some of these patterns in Rails for
instance, like when a method checks its parameter formats or valid
options in the trailing hash. However, overall I'll have to echo
Rick's sentiment.  You don't need something external to enforce an
interface -- if your code is calling something it shouldn't, it's
wrong anyway and you should rightly get a NameError or
NotImplementedError.

Sean

On Sun, Mar 6, 2011 at 9:54 AM, Trent Albright <trent.albright at gmail.com> wrote:
> I've written plenty of production code in Java. I prefer by far duck typing
> in Ruby than having to declare and fulfill interface contracts:
>
> In duck typing, one is concerned with just those aspects of an object that
> are used, rather than with the type of the object itself.
>
> It feels like interfaces in Java are about compile time checking and
> something that the VM runtime cares about. From a developers perspective,
> seeing "implements MyInterface" in Java lets me know what kind of behavior
> the class has or that I should provide. It's an arbitrary contract that
> needs to be fulfilled.?But I think all we really care about is whether a
> given object can receive a given message. We don't need interfaces in Ruby
> to accomplish that.
> In general, in Ruby when we want to make sure objects behave as they are
> supposed to we write specs/tests. In cases where you need to check at
> runtime we use "Object#respond_to?"
> I agree completely with Rick and Sandi.
>>
>> You can use the tests you've written for your modules to prove that the
>> classes that include those modules honor the interface.
>>
>> http://relishapp.com/rspec/rspec-core/v/2-0/dir/example-groups/shared-example-group
>> In your case you might have a contributor_spec, which defined some
>>
>> shared_examples_for "contributor"
>>
>> and then each class that included the contributor module would assert
>> something like
>>
>> describe Participant
>> ?? it_should_behave_like "contributor"
>> ?? ...
>>
>> The tests?ought?to prove that everything is ok AND provide all of the
>> documentation you need.
>> If they don't, improve the tests.
>> Win-win.
>> Of course, this assumes that you're using rspec ?:-) but surely something
>> similar is possible in other test frameworks.
>> -Sandi
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>

From martin.streicher at gmail.com  Mon Mar  7 20:06:20 2011
From: martin.streicher at gmail.com (Martin Streicher)
Date: Mon, 7 Mar 2011 20:06:20 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
Message-ID: <B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>


Tyler: did you ever post a followup? 

On Nov 30, 2010, at 3:00 PM, Tyler Smart wrote:

> I've decided to go with a combination of 
> Request Log Analyzer https://github.com/wvanbergen/request-log-analyzer
> Oink https://github.com/noahd1/oink
> Memprof  https://github.com/ice799/memprof
> I'll let you guys know how successful it was.
> 
> Tyler
> 
> On Tue, Nov 30, 2010 at 1:28 PM, Donald Ball <donald.ball at gmail.com> wrote:
> On Tue, Nov 30, 2010 at 12:22 PM, Aaron Bedra <aaron at aaronbedra.com> wrote:
> > If you are still on1.8.x REE is the ONLY ruby you should be using. The GC
> > and threading fixes make it a deployable Ruby.  I would not use MRI in
> > production for anything.  If you want the lowdown on what was fixed in REE
> > take a look at Hong and Nihn's talk at google on it.  It is very eye
> > opening.
> 
> Having a sysadmin whose got us stuck on 1.8.6 (!) still, and is
> contemplating a move to MRI 1.8.7 instead of REE (or 1.9.2),
> apparently due to some bad blood between the Redhat guys and the REE
> guys, I'd appreciate a link to said talk if you happen to have one, I
> couldn't find it just now. Thanks.
> 
> - donald
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
> 
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110307/f6fb0403/attachment.html>

From tesmar at statsheet.com  Tue Mar  8 07:32:37 2011
From: tesmar at statsheet.com (Tyler Smart)
Date: Tue, 8 Mar 2011 07:32:37 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
Message-ID: <AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>

Well, for the moment we have paused our profiling in order to write more
code, but I was able to use bleakhouse to trackdown  a bad activerecord
optimizer plugin. I removed it and memory leaks went away, some of them :)

RIght now since our rails app has 600 models and will soon be approaching
700-900 with the addition of new sports, I am working on a way to
selectively deploy different sports as different apps using Rails engines. I
want to create whitelists for which files will get deployed as which kind of
server. For ex: if we are deploying an mlb app we will only include the base
models + mlb for a total of 150 or so, not 600. I am hoping that this will
speed the app up. And, we still have to switch to 1.9.2 or JRuby.

On Mon, Mar 7, 2011 at 8:06 PM, Martin Streicher <martin.streicher at gmail.com
> wrote:

>
> Tyler: did you ever post a followup?
>
> On Nov 30, 2010, at 3:00 PM, Tyler Smart wrote:
>
> I've decided to go with a combination of
>
>    - Request Log Analyzer
>    https://github.com/wvanbergen/request-log-analyzer
>    - Oink https://github.com/noahd1/oink
>    - Memprof  https://github.com/ice799/memprof
>
> I'll let you guys know how successful it was.
>
> Tyler
>
> On Tue, Nov 30, 2010 at 1:28 PM, Donald Ball <donald.ball at gmail.com>wrote:
>
>> On Tue, Nov 30, 2010 at 12:22 PM, Aaron Bedra <aaron at aaronbedra.com>
>> wrote:
>> > If you are still on1.8.x REE is the ONLY ruby you should be using. The
>> GC
>> > and threading fixes make it a deployable Ruby.  I would not use MRI in
>> > production for anything.  If you want the lowdown on what was fixed in
>> REE
>> > take a look at Hong and Nihn's talk at google on it.  It is very eye
>> > opening.
>>
>> Having a sysadmin whose got us stuck on 1.8.6 (!) still, and is
>> contemplating a move to MRI 1.8.7 instead of REE (or 1.9.2),
>> apparently due to some bad blood between the Redhat guys and the REE
>> guys, I'd appreciate a link to said talk if you happen to have one, I
>> couldn't find it just now. Thanks.
>>
>> - donald
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/124f6d2e/attachment.html>

From info at lojic.com  Tue Mar  8 10:47:39 2011
From: info at lojic.com (Brian Adkins)
Date: Tue, 8 Mar 2011 10:47:39 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
Message-ID: <B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>

Tyler:

I'm curious about the large number of models you have. Would you mind  
sharing some examples of what drives the need for a new model vs.  
parameterizing an existing model with data when new sports are  
introduced? I'm very unfamiliar with the domain, but at first glance  
the number of models seemed a bit extreme :)

Thanks,
Brian

--
Brian Adkins
Lojic Technologies, LLC
http://lojic.com/




On Mar 8, 2011, at 7:32 AM, Tyler Smart wrote:

> Well, for the moment we have paused our profiling in order to write  
> more code, but I was able to use bleakhouse to trackdown  a bad  
> activerecord optimizer plugin. I removed it and memory leaks went  
> away, some of them :)
>
> RIght now since our rails app has 600 models and will soon be  
> approaching 700-900 with the addition of new sports, I am working on  
> a way to selectively deploy different sports as different apps using  
> Rails engines. I want to create whitelists for which files will get  
> deployed as which kind of server. For ex: if we are deploying an mlb  
> app we will only include the base models + mlb for a total of 150 or  
> so, not 600. I am hoping that this will speed the app up. And, we  
> still have to switch to 1.9.2 or JRuby.
>
> On Mon, Mar 7, 2011 at 8:06 PM, Martin Streicher <martin.streicher at gmail.com 
> > wrote:
>
> Tyler: did you ever post a followup?
>
> On Nov 30, 2010, at 3:00 PM, Tyler Smart wrote:
>
>> I've decided to go with a combination of
>> Request Log Analyzer https://github.com/wvanbergen/request-log-analyzer
>> Oink https://github.com/noahd1/oink
>> Memprof  https://github.com/ice799/memprof
>> I'll let you guys know how successful it was.
>>
>> Tyler
>>
>> On Tue, Nov 30, 2010 at 1:28 PM, Donald Ball  
>> <donald.ball at gmail.com> wrote:
>> On Tue, Nov 30, 2010 at 12:22 PM, Aaron Bedra  
>> <aaron at aaronbedra.com> wrote:
>> > If you are still on1.8.x REE is the ONLY ruby you should be  
>> using. The GC
>> > and threading fixes make it a deployable Ruby.  I would not use  
>> MRI in
>> > production for anything.  If you want the lowdown on what was  
>> fixed in REE
>> > take a look at Hong and Nihn's talk at google on it.  It is very  
>> eye
>> > opening.
>>
>> Having a sysadmin whose got us stuck on 1.8.6 (!) still, and is
>> contemplating a move to MRI 1.8.7 instead of REE (or 1.9.2),
>> apparently due to some bad blood between the Redhat guys and the REE
>> guys, I'd appreciate a link to said talk if you happen to have one, I
>> couldn't find it just now. Thanks.
>>
>> - donald
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/eacb4a3b/attachment.html>

From tesmar at statsheet.com  Tue Mar  8 10:55:38 2011
From: tesmar at statsheet.com (Tyler Smart)
Date: Tue, 8 Mar 2011 10:55:38 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
Message-ID: <AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>

Brian,

The sports are just different enough that we thought that cramming all the
models and writing super-methods which were really ugly would not be worth
the maintenance. The code is somewhat similar, but branches enough that we
thought about separating it. It is an extreme number of models, but
considering the differences between the sports it will only grow with time.

On Tue, Mar 8, 2011 at 10:47 AM, Brian Adkins <info at lojic.com> wrote:

> Tyler:
>
> I'm curious about the large number of models you have. Would you mind
> sharing some examples of what drives the need for a new model vs.
> parameterizing an existing model with data when new sports are introduced?
> I'm very unfamiliar with the domain, but at first glance the number of
> models seemed a bit extreme :)
>
> Thanks,
> Brian
>
> --
> Brian Adkins
> Lojic Technologies, LLC
> http://lojic.com/
>
>
>
>
> On Mar 8, 2011, at 7:32 AM, Tyler Smart wrote:
>
> Well, for the moment we have paused our profiling in order to write more
> code, but I was able to use bleakhouse to trackdown  a bad activerecord
> optimizer plugin. I removed it and memory leaks went away, some of them :)
>
> RIght now since our rails app has 600 models and will soon be approaching
> 700-900 with the addition of new sports, I am working on a way to
> selectively deploy different sports as different apps using Rails engines. I
> want to create whitelists for which files will get deployed as which kind of
> server. For ex: if we are deploying an mlb app we will only include the base
> models + mlb for a total of 150 or so, not 600. I am hoping that this will
> speed the app up. And, we still have to switch to 1.9.2 or JRuby.
>
> On Mon, Mar 7, 2011 at 8:06 PM, Martin Streicher <
> martin.streicher at gmail.com> wrote:
>
>>
>> Tyler: did you ever post a followup?
>>
>> On Nov 30, 2010, at 3:00 PM, Tyler Smart wrote:
>>
>> I've decided to go with a combination of
>>
>>    - Request Log Analyzer
>>    https://github.com/wvanbergen/request-log-analyzer
>>    - Oink https://github.com/noahd1/oink
>>    - Memprof  https://github.com/ice799/memprof
>>
>> I'll let you guys know how successful it was.
>>
>> Tyler
>>
>> On Tue, Nov 30, 2010 at 1:28 PM, Donald Ball <donald.ball at gmail.com>wrote:
>>
>>> On Tue, Nov 30, 2010 at 12:22 PM, Aaron Bedra <aaron at aaronbedra.com>
>>> wrote:
>>> > If you are still on1.8.x REE is the ONLY ruby you should be using. The
>>> GC
>>> > and threading fixes make it a deployable Ruby.  I would not use MRI in
>>> > production for anything.  If you want the lowdown on what was fixed in
>>> REE
>>> > take a look at Hong and Nihn's talk at google on it.  It is very eye
>>> > opening.
>>>
>>> Having a sysadmin whose got us stuck on 1.8.6 (!) still, and is
>>> contemplating a move to MRI 1.8.7 instead of REE (or 1.9.2),
>>> apparently due to some bad blood between the Redhat guys and the REE
>>> guys, I'd appreciate a link to said talk if you happen to have one, I
>>> couldn't find it just now. Thanks.
>>>
>>> - donald
>>> _______________________________________________
>>> raleigh-rb-members mailing list
>>> raleigh-rb-members at rubyforge.org
>>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>>
>>
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>
>>
>>
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/0afbc1e2/attachment-0001.html>

From nathaniel at talbott.ws  Tue Mar  8 11:00:21 2011
From: nathaniel at talbott.ws (Nathaniel Talbott)
Date: Tue, 8 Mar 2011 11:00:21 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
Message-ID: <AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>

On Tue, Mar 8, 2011 at 10:55 AM, Tyler Smart <tesmar at statsheet.com> wrote:

> The sports are just different enough that we thought that cramming all the
> models and writing super-methods which were really ugly would not be worth
> the maintenance. The code is somewhat similar, but branches enough that we
> thought about separating it. It is an extreme number of models, but
> considering the differences between the sports it will only grow with time.

Do you have a different database table for each of those models?


-- 
Nathaniel Talbott
<:((><

From trent.albright at gmail.com  Tue Mar  8 11:13:32 2011
From: trent.albright at gmail.com (Trent Albright)
Date: Tue, 8 Mar 2011 11:13:32 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
Message-ID: <AANLkTikT8pHdNaoA1Ff=tZvpJv9fXWh182g1JtD7ffkA@mail.gmail.com>

I've worked with schemas before that have hundreds of models. Even over 600.
I realize it is certainly not typical but some domains require it.

On Tue, Mar 8, 2011 at 10:47 AM, Brian Adkins <info at lojic.com> wrote:

> Tyler:
>
> I'm curious about the large number of models you have. Would you mind
> sharing some examples of what drives the need for a new model vs.
> parameterizing an existing model with data when new sports are introduced?
> I'm very unfamiliar with the domain, but at first glance the number of
> models seemed a bit extreme :)
>
> Thanks,
> Brian
>
> --
> Brian Adkins
> Lojic Technologies, LLC
> http://lojic.com/
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/19d28036/attachment.html>

From tesmar at statsheet.com  Tue Mar  8 11:57:20 2011
From: tesmar at statsheet.com (Tyler Smart)
Date: Tue, 8 Mar 2011 11:57:20 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
Message-ID: <AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>

Most of them do, about 85%. There is some cleanup that can be done, and I
have a project underway to do that, but that still leaves a significant
number of models, something that will only grow.

On Tue, Mar 8, 2011 at 11:00 AM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:

> On Tue, Mar 8, 2011 at 10:55 AM, Tyler Smart <tesmar at statsheet.com> wrote:
>
> > The sports are just different enough that we thought that cramming all
> the
> > models and writing super-methods which were really ugly would not be
> worth
> > the maintenance. The code is somewhat similar, but branches enough that
> we
> > thought about separating it. It is an extreme number of models, but
> > considering the differences between the sports it will only grow with
> time.
>
> Do you have a different database table for each of those models?
>
>
> --
> Nathaniel Talbott
> <:((><
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/641b686c/attachment.html>

From nathaniel at talbott.ws  Tue Mar  8 12:27:56 2011
From: nathaniel at talbott.ws (Nathaniel Talbott)
Date: Tue, 8 Mar 2011 12:27:56 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
Message-ID: <AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>

On Tue, Mar 8, 2011 at 11:57 AM, Tyler Smart <tesmar at statsheet.com> wrote:

> Most of them do, about 85%. There is some cleanup that can be done, and I
> have a project underway to do that, but that still leaves a significant
> number of models, something that will only grow.

While I concede that there are cases where that many models make
sense, I've found in many cases there's a fundamental assumption that
can be tweaked to simplify drastically. With Spreedly for instance we
went from having one database table per payment gateway to a single
payment gateway table by making one seemingly trivial change. I guess
my next question is: do the models vary primarily in the data they
contain, the behavior they need, some combo, or something else
entirely?

Oh, one other question while I'm at it - how many models are you
adding per sport at this point?


-- 
Nathaniel Talbott
<:((><

From info at lojic.com  Tue Mar  8 12:45:20 2011
From: info at lojic.com (Brian Adkins)
Date: Tue, 8 Mar 2011 12:45:20 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTikT8pHdNaoA1Ff=tZvpJv9fXWh182g1JtD7ffkA@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTikT8pHdNaoA1Ff=tZvpJv9fXWh182g1JtD7ffkA@mail.gmail.com>
Message-ID: <A17EFCE7-5E52-43BE-B60C-9631F744E3B4@lojic.com>

Just to be clear, my point wasn't to imply that using hundreds of  
models was wrong since I have little info on the requirements. I was  
genuinely curious about what requirements might indicate a need for  
hundreds of models. There are many ways to skin the cat :)

With respect to some domains "requiring" that many; I expect that may  
be true, but I also think that the number of models is subjective and  
some folks would use less, some more. One of the things I like about  
Ruby is the flexibility it offers in structuring solutions - class  
inheritance, modules, metaprogramming, dynamic code loading, etc. I've  
been experimenting with using a more functional approach vs. object  
oriented approach in some of my Ruby code, so I was curious about this  
example which seemed to be heavy on the OO side.

For example, given the overly simplistic example below, some people  
might lean toward the two model solution, others might lean toward the  
one model solution, or to an entirely different mechanism. As more  
functionality is added, it will likely become clear which is more  
advantageous for the particular system. For this simple example, it's  
moot.

class Cat
   def speak
     'meow'
   end
end

class Dog
   def speak
     'bark'
   end
end

class Animal
   def initialize speech
     @speech = speech
   end

   def speak
     @speech
   end
end

[ Cat.new,
   Dog.new,
   Animal.new('meow'),
   Animal.new('bark') ].each {|animal| puts animal.speak }

Thanks,
Brian

P.S. why is speak spelled with "ea" and speech with "ee" ? :)

On Mar 8, 2011, at 11:13 AM, Trent Albright wrote:

> I've worked with schemas before that have hundreds of models. Even  
> over 600. I realize it is certainly not typical but some domains  
> require it.
>
> On Tue, Mar 8, 2011 at 10:47 AM, Brian Adkins <info at lojic.com> wrote:
> Tyler:
>
> I'm curious about the large number of models you have. Would you  
> mind sharing some examples of what drives the need for a new model  
> vs. parameterizing an existing model with data when new sports are  
> introduced? I'm very unfamiliar with the domain, but at first glance  
> the number of models seemed a bit extreme :)
>
> Thanks,
> Brian
>
> --
> Brian Adkins
> Lojic Technologies, LLC
> http://lojic.com/
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/e5345087/attachment.html>

From tesmar at statsheet.com  Tue Mar  8 13:42:05 2011
From: tesmar at statsheet.com (Tyler Smart)
Date: Tue, 8 Mar 2011 13:42:05 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
	<AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
Message-ID: <AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>

They vary mostly in the behavior, so I suppose there could be a lib with a
ton of logic in it that does crazy branching. We are moving to a new stats
model that will take a lot of the tables out once we clean up and so for
existing sports I expect the models to go down to about 400, maybe even 350.
If we reworked the code to be sport-agnostic, then we are talking maybe 150
models? Maybe. But that would require a ton of rewriting, something we don't
have time to do and would introduce complexity into the code. At this point
I am thinking it would be easier to segment the app at deploy time.


On Tue, Mar 8, 2011 at 12:27 PM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:

> On Tue, Mar 8, 2011 at 11:57 AM, Tyler Smart <tesmar at statsheet.com> wrote:
>
> > Most of them do, about 85%. There is some cleanup that can be done, and I
> > have a project underway to do that, but that still leaves a significant
> > number of models, something that will only grow.
>
> While I concede that there are cases where that many models make
> sense, I've found in many cases there's a fundamental assumption that
> can be tweaked to simplify drastically. With Spreedly for instance we
> went from having one database table per payment gateway to a single
> payment gateway table by making one seemingly trivial change. I guess
> my next question is: do the models vary primarily in the data they
> contain, the behavior they need, some combo, or something else
> entirely?
>
> Oh, one other question while I'm at it - how many models are you
> adding per sport at this point?
>
>
> --
> Nathaniel Talbott
> <:((><
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/cfba5d56/attachment-0001.html>

From nathaniel at talbott.ws  Tue Mar  8 15:35:15 2011
From: nathaniel at talbott.ws (Nathaniel Talbott)
Date: Tue, 8 Mar 2011 15:35:15 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
	<AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
	<AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>
Message-ID: <AANLkTinM_qfqG5NkDD+CQ_RbQoDrnBsqT+QfTiLLnUoj@mail.gmail.com>

On Tue, Mar 8, 2011 at 1:42 PM, Tyler Smart <tesmar at statsheet.com> wrote:

> They vary mostly in the behavior, so I suppose there could be a lib with a
> ton of logic in it that does crazy branching. We are moving to a new stats
> model that will take a lot of the tables out once we clean up and so for
> existing sports I expect the models to go down to about 400, maybe even 350.
> If we reworked the code to be sport-agnostic, then we are talking maybe 150
> models? Maybe. But that would require a ton of rewriting, something we don't
> have time to do and would introduce complexity into the code. At this point
> I am thinking it would be easier to segment the app at deploy time.

As with Brian, I'm mostly just curious, seeing as you're pushing the
boundaries and edge cases are fun :-)

How many models do you add to the system at this point when a new
sport is added?


-- 
Nathaniel Talbott
<:((><

From tesmar at statsheet.com  Tue Mar  8 18:08:45 2011
From: tesmar at statsheet.com (Tyler Smart)
Date: Tue, 8 Mar 2011 18:08:45 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTinM_qfqG5NkDD+CQ_RbQoDrnBsqT+QfTiLLnUoj@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
	<AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
	<AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>
	<AANLkTinM_qfqG5NkDD+CQ_RbQoDrnBsqT+QfTiLLnUoj@mail.gmail.com>
Message-ID: <AANLkTikDAJKOj6ZvzjYS7_fKniSjpnUB1O66jx5XrypJ@mail.gmail.com>

On average we add 30-40 models per sport. We will be adding 3-4 more sports,
soon as well as a bunch of new functionality which will add even MORE
models.

On Tue, Mar 8, 2011 at 3:35 PM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:

> On Tue, Mar 8, 2011 at 1:42 PM, Tyler Smart <tesmar at statsheet.com> wrote:
>
> > They vary mostly in the behavior, so I suppose there could be a lib with
> a
> > ton of logic in it that does crazy branching. We are moving to a new
> stats
> > model that will take a lot of the tables out once we clean up and so for
> > existing sports I expect the models to go down to about 400, maybe even
> 350.
> > If we reworked the code to be sport-agnostic, then we are talking maybe
> 150
> > models? Maybe. But that would require a ton of rewriting, something we
> don't
> > have time to do and would introduce complexity into the code. At this
> point
> > I am thinking it would be easier to segment the app at deploy time.
>
> As with Brian, I'm mostly just curious, seeing as you're pushing the
> boundaries and edge cases are fun :-)
>
> How many models do you add to the system at this point when a new
> sport is added?
>
>
> --
> Nathaniel Talbott
> <:((><
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/b2b0a04d/attachment.html>

From jrubyforge-raleigh-rb at indythinker.com  Tue Mar  8 15:39:33 2011
From: jrubyforge-raleigh-rb at indythinker.com (Justis Peters)
Date: Tue, 08 Mar 2011 15:39:33 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <4D76933F.3000900@indythinker.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>	<AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
	<AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>
	<4D76933F.3000900@indythinker.com>
Message-ID: <4D769405.7090606@indythinker.com>

This was originally sent from the wrong address, so I am resending it:

On 03/08/2011 03:36 PM, Justis Peters wrote:
> Tyler,
>
> How many rows do you expect to have in these tables? If it's a lot, 
> then the number of tables you have may have a significant impact on 
> index performance. If it will be rare that you want to join tables 
> across multiple sports, you will probably get the best performance by 
> keeping them in separate tables.
>
> Alternately, your database might support "table partitioning" or even 
> "index partitioning", which could solve the issue of index 
> performance. Here's some relevant info for PostgreSQL and MySQL (IIRC, 
> there is also support in Oracle and MS SQL Server) :
> http://www.postgresql.org/docs/9.0/static/ddl-partitioning.html
> http://dev.mysql.com/doc/refman/5.1/en/partitioning.html
>
> You might also want to consider how many queries it takes to 
> instantiate a model from the database. Some of the solutions that have 
> been suggested here would involve many iterations between rails and 
> the database before you would be done instantiating a collection.
>
> Kind regards,
> Justis
>
> On 03/08/2011 01:42 PM, Tyler Smart wrote:
>> They vary mostly in the behavior, so I suppose there could be a lib 
>> with a ton of logic in it that does crazy branching. We are moving to 
>> a new stats model that will take a lot of the tables out once we 
>> clean up and so for existing sports I expect the models to go down to 
>> about 400, maybe even 350. If we reworked the code to be 
>> sport-agnostic, then we are talking maybe 150 models? Maybe. But that 
>> would require a ton of rewriting, something we don't have time to do 
>> and would introduce complexity into the code. At this point I am 
>> thinking it would be easier to segment the app at deploy time.
>>
>> On Tue, Mar 8, 2011 at 12:27 PM, Nathaniel Talbott 
>> <nathaniel at talbott.ws <mailto:nathaniel at talbott.ws>> wrote:
>>
>>     On Tue, Mar 8, 2011 at 11:57 AM, Tyler Smart
>>     <tesmar at statsheet.com <mailto:tesmar at statsheet.com>> wrote:
>>
>>     > Most of them do, about 85%. There is some cleanup that can be
>>     done, and I
>>     > have a project underway to do that, but that still leaves a
>>     significant
>>     > number of models, something that will only grow.
>>
>>     While I concede that there are cases where that many models make
>>     sense, I've found in many cases there's a fundamental assumption that
>>     can be tweaked to simplify drastically. With Spreedly for instance we
>>     went from having one database table per payment gateway to a single
>>     payment gateway table by making one seemingly trivial change. I guess
>>     my next question is: do the models vary primarily in the data they
>>     contain, the behavior they need, some combo, or something else
>>     entirely?
>>
>>     Oh, one other question while I'm at it - how many models are you
>>     adding per sport at this point?
>>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110308/6404ef75/attachment.html>

From tesmar at statsheet.com  Wed Mar  9 07:56:27 2011
From: tesmar at statsheet.com (Tyler Smart)
Date: Wed, 9 Mar 2011 07:56:27 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <4D769405.7090606@indythinker.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
	<AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
	<AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>
	<4D76933F.3000900@indythinker.com> <4D769405.7090606@indythinker.com>
Message-ID: <AANLkTin8GLZxaCgdR0p9EK0BXHrtj0JvSzSni1d-HPUW@mail.gmail.com>

It depends on the sport. Some stat tables like mlb have upwards of 1 - 2
billion rows (keeps growing). Nba has only in the neighborhood of 100-200
million. A lot of the smaller tables only 1 or 2 million or so, and the
smallest a few thousand. We are partitioning and using indexes wisely.
Separating them into sports was done for performance, and it has worked
nicely. I think this is largely an issue of  "OH NO I HAVE TOO MANY MODELS"
and I'll keep you up to date with how the deploy-time solution works.

On Tue, Mar 8, 2011 at 3:39 PM, Justis Peters <
jrubyforge-raleigh-rb at indythinker.com> wrote:

>  This was originally sent from the wrong address, so I am resending it:
>
> On 03/08/2011 03:36 PM, Justis Peters wrote:
>
> Tyler,
>
> How many rows do you expect to have in these tables? If it's a lot, then
> the number of tables you have may have a significant impact on index
> performance. If it will be rare that you want to join tables across multiple
> sports, you will probably get the best performance by keeping them in
> separate tables.
>
> Alternately, your database might support "table partitioning" or even
> "index partitioning", which could solve the issue of index performance.
> Here's some relevant info for PostgreSQL and MySQL (IIRC, there is also
> support in Oracle and MS SQL Server) :
>   http://www.postgresql.org/docs/9.0/static/ddl-partitioning.html
>   http://dev.mysql.com/doc/refman/5.1/en/partitioning.html
>
> You might also want to consider how many queries it takes to instantiate a
> model from the database. Some of the solutions that have been suggested here
> would involve many iterations between rails and the database before you
> would be done instantiating a collection.
>
> Kind regards,
> Justis
>
>
> On 03/08/2011 01:42 PM, Tyler Smart wrote:
>
> They vary mostly in the behavior, so I suppose there could be a lib with a
> ton of logic in it that does crazy branching. We are moving to a new stats
> model that will take a lot of the tables out once we clean up and so for
> existing sports I expect the models to go down to about 400, maybe even 350.
> If we reworked the code to be sport-agnostic, then we are talking maybe 150
> models? Maybe. But that would require a ton of rewriting, something we don't
> have time to do and would introduce complexity into the code. At this point
> I am thinking it would be easier to segment the app at deploy time.
>
> On Tue, Mar 8, 2011 at 12:27 PM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:
>
>> On Tue, Mar 8, 2011 at 11:57 AM, Tyler Smart <tesmar at statsheet.com>
>> wrote:
>>
>> > Most of them do, about 85%. There is some cleanup that can be done, and
>> I
>> > have a project underway to do that, but that still leaves a significant
>> > number of models, something that will only grow.
>>
>>  While I concede that there are cases where that many models make
>> sense, I've found in many cases there's a fundamental assumption that
>> can be tweaked to simplify drastically. With Spreedly for instance we
>> went from having one database table per payment gateway to a single
>> payment gateway table by making one seemingly trivial change. I guess
>> my next question is: do the models vary primarily in the data they
>> contain, the behavior they need, some combo, or something else
>> entirely?
>>
>> Oh, one other question while I'm at it - how many models are you
>> adding per sport at this point?
>>
>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110309/3e2476f6/attachment-0001.html>

From nathaniel at talbott.ws  Wed Mar  9 09:31:30 2011
From: nathaniel at talbott.ws (Nathaniel Talbott)
Date: Wed, 9 Mar 2011 09:31:30 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTikDAJKOj6ZvzjYS7_fKniSjpnUB1O66jx5XrypJ@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTikOCn0Jk7nShSYy8QHgaxC07mbu1_iDZPx0qtNf@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
	<AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
	<AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>
	<AANLkTinM_qfqG5NkDD+CQ_RbQoDrnBsqT+QfTiLLnUoj@mail.gmail.com>
	<AANLkTikDAJKOj6ZvzjYS7_fKniSjpnUB1O66jx5XrypJ@mail.gmail.com>
Message-ID: <AANLkTi=csuyL9ioLA_HX9EwOa4gn_bezT6fUSsg1K==9@mail.gmail.com>

On Tue, Mar 8, 2011 at 6:08 PM, Tyler Smart <tesmar at statsheet.com> wrote:

> On average we add 30-40 models per sport. We will be adding 3-4 more sports,
> soon as well as a bunch of new functionality which will add even MORE
> models.

Wow, that's about how many models the average Rails app has total
after a year or two of life; Spreedly Subscriptions is at 100 now
after three years. Of course you're dealing with an edge case - most
apps don't have tables with billions of rows in them until they've
been around for years, either - so I wouldn't be overly concerned.
Your thought of partitioning the models somehow at deploy is
interesting, and I'm curious to hear how it goes.

FYI, if I was looking at refactoring this somehow, I'd probably start
by looking not at finding commonalities between sports, but rather at
finding a way to cut down on the number of models/tables needed on a
per sport basis. I'd also probably be looking hard at the usage models
to see if a non-relational database of some type (Mongo, Redis, Riak,
etc.) would be a better fit for a portion of the problem and thereby
reduce complexity.

Thanks for sharing Tyler - it's great to get a peek into what ya'll
are up to over there at Statsheet. In a few months we might have to
have you come in to Raleigh.rb and do a talk on what you've learned
working on a such a cool and unique problem.


-- 
Nathaniel Talbott
<:((><

From aok at chariotsolutions.com  Wed Mar  9 10:03:26 2011
From: aok at chariotsolutions.com (Andrea O. K. Wright)
Date: Wed, 9 Mar 2011 10:03:26 -0500
Subject: [raleigh.rb] Philly Emerging Tech Conference User Group Discount
Message-ID: <AANLkTinG+mVQYLTZLm-bd=s9AqfKGMe+__aMwidMiaBq@mail.gmail.com>

Earlier this year, I emailed about the annual Philly Emerging Tech
Conference (a.k.a. "Emerging Technologies for the Enterprise" and "Philly
ETE"), which is set for April 27 and April 28.

Since then, the session details have been posted on the conference website:
http://phillyemergingtech.com/2011/sessions

Also, a discount is now available for user group members. With the $25 user
group discount (use discount code "ruby1"), a pass for this 2-day, 5-track
event costs $340.

The per-ticket cost goes down to $255 if you can take advantage of the "4
for the price of 3" rate (4 colleagues or friends, not necessarily from the
same company, registering at the same time) on top of the user group
discount.

To register: http://phillyemergingtech.com/2011/register

Talks by Rubyists include:

* David A. Black (author of The Well-Grounded Rubyist)
   -- Command Performance: the why and whether of small-scale optimizations

* Ola Bini (JRuby Core Team)
   -- Polyglot Patterns

* David Chelimsky (RSpec Lead Developer)
   -- Developer eXperience: The Forgotten User

* Davis Frank (contributor to the Jasmine BDD/JS framework)
   -- Practical Jasmine: Test-Driving Your JavaScript with BDD

* Jay McGavren (maintainer, Jemini JRuby game framework)
   -- Ruby on Android with Ruboto

* Yehuda Katz (SproutCore project team; Rails Core team)
   -- An App Within an App: Rails 3 Engines, and What's Coming in Rails 3.1

* Nick Sieger (JRuby Core Team)
   -- Transitioning Legacy Java to Rails with JRuby: Step By Step

* Bruce Tate (author of Seven Languages in Seven Weeks)
   -- Mary Poppins Meets the Matrix: Seven Languages at the Cinema

* Jim Weirich (creator of the Ruby tool, rake)
   -- Are You Satisfied with Your Tests?

The conference features a wide array of technology topics, in addition to
Ruby-related sessions. Here's a sampling from the program:

* Jonas Boner (Akka Framework Creator)
   -- Akka: Simpler Scalability, Fault-Tolerance, Concurrency & Remoting
Through Actors

* Ryan Dahl (creator of Node.js)
   -- Distributed Systems with Node.js

* Debasish Ghosh (author of DSLs in Action)
   -- DSLs: Does Expressiveness for Domain Experts Have to Compromise the
Underlying Implementation

* Ron Jeffries & Chet Hendrickson (co-authors of The Agile Manifesto)
   -- A Retrospective Rant: 15 Years in the "Agile" Business

* Dave Johnson (CTO & co-founder of Nitobi, creators of PhoneGap)
   -- PhoneGap: Past, Present & Future

* David Kaneda (creator of jQtouch)
   -- Building Rich User Experiences with Sencha Touch

* Stuart Sierra (Clojure/core team)
   -- Clojure: Lisp for the Real Work

* Kevin Smith (author, Erlang in Practice screencasts)
   -- Get some REST with Erlang: Creating Rock-solid Web Services with
Webmachine
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110309/32ea902a/attachment.html>

From info at lojic.com  Wed Mar  9 12:48:31 2011
From: info at lojic.com (Brian Adkins)
Date: Wed, 9 Mar 2011 12:48:31 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <AANLkTin8GLZxaCgdR0p9EK0BXHrtj0JvSzSni1d-HPUW@mail.gmail.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
	<AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
	<AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>
	<4D76933F.3000900@indythinker.com>
	<4D769405.7090606@indythinker.com> 
	<AANLkTin8GLZxaCgdR0p9EK0BXHrtj0JvSzSni1d-HPUW@mail.gmail.com>
Message-ID: <0A2B9940-AF03-453F-B342-FF4027F18C62@lojic.com>

2 billion rows ?! Do you mind sharing what database you're using, the  
amount of RAM consumed and how large the indexes are for the mlb table?

I recently developed a Rails/MongoDB app with 40+ million rows, and  
the indexes (after some painful optimization) still take about 10GB.  
It runs fine on a 16GB server, but we'd have some serious performance  
issues with 2 billion rows :) For the type of queries we're running  
MongoDB *really* wants the indexes to fit in RAM.

In case anyone's using MongoDB, here is the output from db.stats();

 > db.stats();
{
	"collections" : 6,
	"objects" : 42201870,
	"avgObjSize" : 352.8811947906574,
	"dataSize" : 14892246308,
	"storageSize" : 15890273024,
	"numExtents" : 44,
	"indexes" : 10,
	"indexSize" : 10662621888,
	"fileSize" : 36423335936,
	"ok" : 1
}


Thanks,
Brian

On Mar 9, 2011, at 7:56 AM, Tyler Smart wrote:

> It depends on the sport. Some stat tables like mlb have upwards of 1  
> - 2 billion rows (keeps growing). Nba has only in the neighborhood  
> of 100-200 million. A lot of the smaller tables only 1 or 2 million  
> or so, and the smallest a few thousand. We are partitioning and  
> using indexes wisely. Separating them into sports was done for  
> performance, and it has worked nicely. I think this is largely an  
> issue of  "OH NO I HAVE TOO MANY MODELS" and I'll keep you up to  
> date with how the deploy-time solution works.
>
> On Tue, Mar 8, 2011 at 3:39 PM, Justis Peters <jrubyforge-raleigh-rb at indythinker.com 
> > wrote:
> This was originally sent from the wrong address, so I am resending it:
>
> On 03/08/2011 03:36 PM, Justis Peters wrote:
>>
>> Tyler,
>>
>> How many rows do you expect to have in these tables? If it's a lot,  
>> then the number of tables you have may have a significant impact on  
>> index performance. If it will be rare that you want to join tables  
>> across multiple sports, you will probably get the best performance  
>> by keeping them in separate tables.
>>
>> Alternately, your database might support "table partitioning" or  
>> even "index partitioning", which could solve the issue of index  
>> performance. Here's some relevant info for PostgreSQL and MySQL  
>> (IIRC, there is also support in Oracle and MS SQL Server) :
>>   http://www.postgresql.org/docs/9.0/static/ddl-partitioning.html
>>   http://dev.mysql.com/doc/refman/5.1/en/partitioning.html
>>
>> You might also want to consider how many queries it takes to  
>> instantiate a model from the database. Some of the solutions that  
>> have been suggested here would involve many iterations between  
>> rails and the database before you would be done instantiating a  
>> collection.
>>
>> Kind regards,
>> Justis
>>
>>
>> On 03/08/2011 01:42 PM, Tyler Smart wrote:
>>>
>>> They vary mostly in the behavior, so I suppose there could be a  
>>> lib with a ton of logic in it that does crazy branching. We are  
>>> moving to a new stats model that will take a lot of the tables out  
>>> once we clean up and so for existing sports I expect the models to  
>>> go down to about 400, maybe even 350. If we reworked the code to  
>>> be sport-agnostic, then we are talking maybe 150 models? Maybe.  
>>> But that would require a ton of rewriting, something we don't have  
>>> time to do and would introduce complexity into the code. At this  
>>> point I am thinking it would be easier to segment the app at  
>>> deploy time.
>>>
>>> On Tue, Mar 8, 2011 at 12:27 PM, Nathaniel Talbott <nathaniel at talbott.ws 
>>> > wrote:
>>> On Tue, Mar 8, 2011 at 11:57 AM, Tyler Smart  
>>> <tesmar at statsheet.com> wrote:
>>>
>>> > Most of them do, about 85%. There is some cleanup that can be  
>>> done, and I
>>> > have a project underway to do that, but that still leaves a  
>>> significant
>>> > number of models, something that will only grow.
>>>
>>> While I concede that there are cases where that many models make
>>> sense, I've found in many cases there's a fundamental assumption  
>>> that
>>> can be tweaked to simplify drastically. With Spreedly for instance  
>>> we
>>> went from having one database table per payment gateway to a single
>>> payment gateway table by making one seemingly trivial change. I  
>>> guess
>>> my next question is: do the models vary primarily in the data they
>>> contain, the behavior they need, some combo, or something else
>>> entirely?
>>>
>>> Oh, one other question while I'm at it - how many models are you
>>> adding per sport at this point?
>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110309/39ce354f/attachment.html>

From tesmar at statsheet.com  Wed Mar  9 14:04:21 2011
From: tesmar at statsheet.com (Tyler Smart)
Date: Wed, 9 Mar 2011 14:04:21 -0500
Subject: [raleigh.rb] Profiling a Rails App
In-Reply-To: <0A2B9940-AF03-453F-B342-FF4027F18C62@lojic.com>
References: <AANLkTim=5StS_nKOGVRoKL1959CkVSEJhwb38HSAwGCK@mail.gmail.com>
	<AANLkTimk3+B3=OJmHanNWXUPSuwwKzELadAJD4wjGWT7@mail.gmail.com>
	<AANLkTikpz+Eq93agSHf8VXgqscu-2DRtbOvePWuzuLXh@mail.gmail.com>
	<AANLkTimwgzbPudYY+AbKzpeASZgvGHDyrAZW_k9=2xxd@mail.gmail.com>
	<4447040F-AA10-47CB-9B38-A38FD819FAE1@aaronbedra.com>
	<AANLkTimVy9i5y9uA7hk_fu-ojc3x5LteSiNp54OTGeOK@mail.gmail.com>
	<AANLkTimP_VDsS=RmESh=3gxDQVVHsvrxcW3JCW62u+X3@mail.gmail.com>
	<B8E85BDE-603C-4CC8-AF28-9AB474C55EE5@gmail.com>
	<AANLkTinK41WsCz41UV7F0EpXrN6Zm7WSF8R7xs+Z=6iA@mail.gmail.com>
	<B8AFFF49-0E5A-4DA3-A02F-E6FE2136106A@lojic.com>
	<AANLkTi=i7t1jqZrbBUqPXvqUVRz4J1_+WJg1SPq0Tpje@mail.gmail.com>
	<AANLkTimX7B8+7AyLA6xhDEjFYtjfnTf9rGF2adQ1mpHp@mail.gmail.com>
	<AANLkTintZfqCZV_p79B3a0oCfQ1tBZGcSV8Hsj8Ms8Up@mail.gmail.com>
	<AANLkTi=watDabsfuy_oxQLW6t3AqUp0fZd5qz_F6N5rs@mail.gmail.com>
	<AANLkTinDJGtYC++MhqY-_XhAq9dE=dgavn-nTfU8Bdtw@mail.gmail.com>
	<4D76933F.3000900@indythinker.com> <4D769405.7090606@indythinker.com>
	<AANLkTin8GLZxaCgdR0p9EK0BXHrtj0JvSzSni1d-HPUW@mail.gmail.com>
	<0A2B9940-AF03-453F-B342-FF4027F18C62@lojic.com>
Message-ID: <AANLkTik8vPJAF2nfBJNkztc4PfPwFtb5jihBofKKR-0Q@mail.gmail.com>

Yes, we are approaching 1 billion in our stats processing now for the last
100 years. We are using Amazon EC2 machines with 8 core CPUs and 7.5GB of
Ram. We are using JRuby to multithread the stats processing. The tables +
indexes are now running 80-100GB, but when we add ranks they will shoot up
to 500-700GB. I think Amazon will let us have 2TB drives.  The indexes take
up most of the RAM, the tables are much smaller. So there is no way to fit a
50GB index much less a 300GB index in RAM.

On Wed, Mar 9, 2011 at 12:48 PM, Brian Adkins <info at lojic.com> wrote:

> 2 billion rows ?! Do you mind sharing what database you're using, the
> amount of RAM consumed and how large the indexes are for the mlb table?
>
> I recently developed a Rails/MongoDB app with 40+ million rows, and the
> indexes (after some painful optimization) still take about 10GB. It runs
> fine on a 16GB server, but we'd have some serious performance issues with 2
> billion rows :) For the type of queries we're running MongoDB *really* wants
> the indexes to fit in RAM.
>
> In case anyone's using MongoDB, here is the output from db.stats();
>
> > db.stats();
> {
> "collections" : 6,
>  "objects" : 42201870,
> "avgObjSize" : 352.8811947906574,
> "dataSize" : 14892246308,
>  "storageSize" : 15890273024,
> "numExtents" : 44,
> "indexes" : 10,
>  "indexSize" : 10662621888,
> "fileSize" : 36423335936,
> "ok" : 1
> }
>
>
> Thanks,
> Brian
>
> On Mar 9, 2011, at 7:56 AM, Tyler Smart wrote:
>
> It depends on the sport. Some stat tables like mlb have upwards of 1 - 2
> billion rows (keeps growing). Nba has only in the neighborhood of 100-200
> million. A lot of the smaller tables only 1 or 2 million or so, and the
> smallest a few thousand. We are partitioning and using indexes wisely.
> Separating them into sports was done for performance, and it has worked
> nicely. I think this is largely an issue of  "OH NO I HAVE TOO MANY MODELS"
> and I'll keep you up to date with how the deploy-time solution works.
>
> On Tue, Mar 8, 2011 at 3:39 PM, Justis Peters <
> jrubyforge-raleigh-rb at indythinker.com> wrote:
>
>>  This was originally sent from the wrong address, so I am resending it:
>>
>> On 03/08/2011 03:36 PM, Justis Peters wrote:
>>
>> Tyler,
>>
>> How many rows do you expect to have in these tables? If it's a lot, then
>> the number of tables you have may have a significant impact on index
>> performance. If it will be rare that you want to join tables across multiple
>> sports, you will probably get the best performance by keeping them in
>> separate tables.
>>
>> Alternately, your database might support "table partitioning" or even
>> "index partitioning", which could solve the issue of index performance.
>> Here's some relevant info for PostgreSQL and MySQL (IIRC, there is also
>> support in Oracle and MS SQL Server) :
>>   http://www.postgresql.org/docs/9.0/static/ddl-partitioning.html
>>   http://dev.mysql.com/doc/refman/5.1/en/partitioning.html
>>
>> You might also want to consider how many queries it takes to instantiate a
>> model from the database. Some of the solutions that have been suggested here
>> would involve many iterations between rails and the database before you
>> would be done instantiating a collection.
>>
>> Kind regards,
>> Justis
>>
>>
>> On 03/08/2011 01:42 PM, Tyler Smart wrote:
>>
>> They vary mostly in the behavior, so I suppose there could be a lib with a
>> ton of logic in it that does crazy branching. We are moving to a new stats
>> model that will take a lot of the tables out once we clean up and so for
>> existing sports I expect the models to go down to about 400, maybe even 350.
>> If we reworked the code to be sport-agnostic, then we are talking maybe 150
>> models? Maybe. But that would require a ton of rewriting, something we don't
>> have time to do and would introduce complexity into the code. At this point
>> I am thinking it would be easier to segment the app at deploy time.
>>
>> On Tue, Mar 8, 2011 at 12:27 PM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:
>>
>>> On Tue, Mar 8, 2011 at 11:57 AM, Tyler Smart <tesmar at statsheet.com>
>>> wrote:
>>>
>>> > Most of them do, about 85%. There is some cleanup that can be done, and
>>> I
>>> > have a project underway to do that, but that still leaves a significant
>>> > number of models, something that will only grow.
>>>
>>>  While I concede that there are cases where that many models make
>>> sense, I've found in many cases there's a fundamental assumption that
>>> can be tweaked to simplify drastically. With Spreedly for instance we
>>> went from having one database table per payment gateway to a single
>>> payment gateway table by making one seemingly trivial change. I guess
>>> my next question is: do the models vary primarily in the data they
>>> contain, the behavior they need, some combo, or something else
>>> entirely?
>>>
>>> Oh, one other question while I'm at it - how many models are you
>>> adding per sport at this point?
>>>
>>
>>
>> _______________________________________________
>> raleigh-rb-members mailing list
>> raleigh-rb-members at rubyforge.org
>> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
>
>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110309/394e2c13/attachment-0001.html>

From nathaniel at talbott.ws  Tue Mar 15 11:13:00 2011
From: nathaniel at talbott.ws (Nathaniel Talbott)
Date: Tue, 15 Mar 2011 11:13:00 -0400
Subject: [raleigh.rb] Pre-meeting Chow
Message-ID: <AANLkTikr0ej1Aeakh6eY_G-36e84JcTB08qtBxtV-hYJ@mail.gmail.com>

Chow tonight will be had at Randy's Pizza:

 http://bit.ly/gafHf7
 http://www.randyspizzartp.com/

I'll be there by 5:30, and everyone is invited to join me to chat a
bit and unwind before heading over to iContact for the meeting proper.

See you there!


-- 
Nathaniel Talbott
<:((><

From tesmar at statsheet.com  Tue Mar 15 15:19:30 2011
From: tesmar at statsheet.com (Tyler Smart)
Date: Tue, 15 Mar 2011 15:19:30 -0400
Subject: [raleigh.rb] Pre-meeting Chow
In-Reply-To: <AANLkTikr0ej1Aeakh6eY_G-36e84JcTB08qtBxtV-hYJ@mail.gmail.com>
References: <AANLkTikr0ej1Aeakh6eY_G-36e84JcTB08qtBxtV-hYJ@mail.gmail.com>
Message-ID: <AANLkTin2aYENUJqA4SbPdPEVBX+FByU9dwO9BueQt+Rw@mail.gmail.com>

See all of you there, seeing as how it is right across the street.

On Tue, Mar 15, 2011 at 11:13 AM, Nathaniel Talbott <nathaniel at talbott.ws>wrote:

> Chow tonight will be had at Randy's Pizza:
>
>  http://bit.ly/gafHf7
>  http://www.randyspizzartp.com/
>
> I'll be there by 5:30, and everyone is invited to join me to chat a
> bit and unwind before heading over to iContact for the meeting proper.
>
> See you there!
>
>
> --
> Nathaniel Talbott
> <:((><
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110315/79e069fc/attachment.html>

From blakewatters at gmail.com  Wed Mar 16 19:52:15 2011
From: blakewatters at gmail.com (Blake Watters)
Date: Wed, 16 Mar 2011 19:52:15 -0400
Subject: [raleigh.rb] RestKit Slides
Message-ID: <AANLkTineevOtOzLxByPvBu3XhdF11LY+uoY1eDrmmhqQ@mail.gmail.com>

Thanks to Nathaniel, iContact, and everyone who came out for last night's
RestKit presentation. As requested, I've thrown the slides up on my Dropbox:
Keynote:
https://www.dropbox.com/s/maog7z0m87a4avv/Two%20Toasters%20-%20Presentation%20-%20RestKit%20for%20Rubyists.key
PDF:
https://www.dropbox.com/s/t5mhs0m5x3se544/Two%20Toasters%20-%20Presentation%20-%20RestKit%20for%20Rubyists.pdf

If anybody is interested in RestKit or needs some help getting started, feel
free to hit me up.

Best,
Blake
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110316/13222740/attachment.html>

From thomas at ravinggenius.com  Sat Mar 19 13:13:46 2011
From: thomas at ravinggenius.com (Thomas Ingram)
Date: Sat, 19 Mar 2011 13:13:46 -0400
Subject: [raleigh.rb] help with dynamically setting root controller#action
Message-ID: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>

I'm building a CMS, and I'm trying to allow a site admin to specify
what they want on the home page. I have some code that hopefully
illustrates my intent (https://gist.github.com/877618). The code is
not currently working, and I'm not sure how to make it work. Has
anyone else ever come across this and found a way to make it work?

-- 
Thomas Ingram ><>

From steve at iannopollo.com  Sat Mar 19 13:43:37 2011
From: steve at iannopollo.com (Steve Iannopollo)
Date: Sat, 19 Mar 2011 13:43:37 -0400
Subject: [raleigh.rb] help with dynamically setting root
	controller#action
In-Reply-To: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>
References: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>
Message-ID: <2DC40691-B938-4BD8-87BE-1D1F4975AE9A@iannopollo.com>

Wouldn't the easiest thing to do is make the root_url go to an action (always the same action), and that action determines what to show?

Your routes shouldn't care, since the root url is always going to be '/', right?

-Steve

On Mar 19, 2011, at 1:13 PM, Thomas Ingram wrote:

> I'm building a CMS, and I'm trying to allow a site admin to specify
> what they want on the home page. I have some code that hopefully
> illustrates my intent (https://gist.github.com/877618). The code is
> not currently working, and I'm not sure how to make it work. Has
> anyone else ever come across this and found a way to make it work?
> 
> -- 
> Thomas Ingram ><>
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members


From thomas at ravinggenius.com  Sat Mar 19 14:33:50 2011
From: thomas at ravinggenius.com (Thomas Ingram)
Date: Sat, 19 Mar 2011 14:33:50 -0400
Subject: [raleigh.rb] help with dynamically setting root
	controller#action
In-Reply-To: <2DC40691-B938-4BD8-87BE-1D1F4975AE9A@iannopollo.com>
References: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>
	<2DC40691-B938-4BD8-87BE-1D1F4975AE9A@iannopollo.com>
Message-ID: <AANLkTimmsi046ADr=Wcy3HHWN-=gwJKVqvdJnhjmS+px@mail.gmail.com>

You are correct that the routes don't much care; I was just thinking I
didn't need a whole controller for this one special action. I updated
the gist to use a controller....

I want to do more than display a different view: I want to execute the
specified action and let that action determine what to do (set ivars,
run before_filters etc) and what to render. If I return the action, it
seems to be ignored.

On Sat, Mar 19, 2011 at 13:43, Steve Iannopollo <steve at iannopollo.com> wrote:
> Wouldn't the easiest thing to do is make the root_url go to an action (always the same action), and that action determines what to show?
>
> Your routes shouldn't care, since the root url is always going to be '/', right?

-- 
Thomas Ingram ><>

From rick.denatale at gmail.com  Sat Mar 19 14:50:23 2011
From: rick.denatale at gmail.com (Rick DeNatale)
Date: Sat, 19 Mar 2011 14:50:23 -0400
Subject: [raleigh.rb] help with dynamically setting root
	controller#action
In-Reply-To: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>
References: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>
Message-ID: <AANLkTi=8MTNT5Cp19hi_opUX81=mhGF5NEL1P6d0aAKk@mail.gmail.com>

On Sat, Mar 19, 2011 at 1:13 PM, Thomas Ingram <thomas at ravinggenius.com> wrote:
> I'm building a CMS, and I'm trying to allow a site admin to specify
> what they want on the home page. I have some code that hopefully
> illustrates my intent (https://gist.github.com/877618). The code is
> not currently working, and I'm not sure how to make it work. Has
> anyone else ever come across this and found a way to make it work?

I'm not sure how to tell you to make this work but..

first of all I assume that this is part of your config/routes.rb file.

second

root :to => dynamic_action

is the same as

match "/", dynamic_action :as => :root

match expects the second argument to either be a string, or something
which can act as a rack endpoint, now your lambda isn't a string, so
it's going to be treated as a rack middleware.

It's going to be called with the rack environment as the parameter,
and it needs to return a 3 element array with the response status,
headers, and body.  Your lambda, returns a string, which ain't gonna
work.


-- 
Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Github: http://github.com/rubyredrick
Twitter: @RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale

From rick.denatale at gmail.com  Sat Mar 19 15:22:29 2011
From: rick.denatale at gmail.com (Rick DeNatale)
Date: Sat, 19 Mar 2011 15:22:29 -0400
Subject: [raleigh.rb] help with dynamically setting root
	controller#action
In-Reply-To: <AANLkTimmsi046ADr=Wcy3HHWN-=gwJKVqvdJnhjmS+px@mail.gmail.com>
References: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>
	<2DC40691-B938-4BD8-87BE-1D1F4975AE9A@iannopollo.com>
	<AANLkTimmsi046ADr=Wcy3HHWN-=gwJKVqvdJnhjmS+px@mail.gmail.com>
Message-ID: <AANLkTikuYdZbfoX3OX7UZTcq6A+kxcTpPU52gN8BT+vE@mail.gmail.com>

On Sat, Mar 19, 2011 at 2:33 PM, Thomas Ingram <thomas at ravinggenius.com> wrote:
> You are correct that the routes don't much care; I was just thinking I
> didn't need a whole controller for this one special action. I updated
> the gist to use a controller....
>
> I want to do more than display a different view: I want to execute the
> specified action and let that action determine what to do (set ivars,
> run before_filters etc) and what to render. If I return the action, it
> seems to be ignored.
>

I think you are mixing the rack and actioncontroller abstraction
levels.  Let's go back to your original gist, something like this
MIGHT be a start:

dynamic_action = lambda do |env|  # I renamed the parameter since
that's the conventional name for a call to a rack middleware
  setting = Setting[:site, :home_page].split '#'
  params ||= {}

  case setting.count
  when 1 # show a specific node
    params[:path] = setting.first
    s_controller, s_action = :nodes, :show_human
  when 2, 3 # show an arbitrary controller#action pair, with optional id
    s_controller, s_action, params[:id] = setting
  else
    # this shouldn't happen...
    # fall back to just listing nodes
    s_controller, s_action = :nodes, :index
  end

  "#{s_controller}_controller".classify.constantize.action(s_action).call(env)
end

root :to => dynamic_action

But this isn't doing the right thing in terms of setting the
parameters in the rack request, you need to read and understand what's
happening in the actionpack routing code in rails 3 to make this work.

If you go this route (pun partially intended) I'm afraid you have a
ways to go.  I think Steve's suggesion of just using a static route
and then having the action invoked for the root url use either
redirects, or calls to other actions (which would probably have to
explicitly name the views they render) is going to work better than
trying to monkey patch rails routing, and will be less likely to break
with new rails releases.

-- 
Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Github: http://github.com/rubyredrick
Twitter: @RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale

From rick.denatale at gmail.com  Sat Mar 19 15:26:04 2011
From: rick.denatale at gmail.com (Rick DeNatale)
Date: Sat, 19 Mar 2011 15:26:04 -0400
Subject: [raleigh.rb] help with dynamically setting root
	controller#action
In-Reply-To: <AANLkTikuYdZbfoX3OX7UZTcq6A+kxcTpPU52gN8BT+vE@mail.gmail.com>
References: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>
	<2DC40691-B938-4BD8-87BE-1D1F4975AE9A@iannopollo.com>
	<AANLkTimmsi046ADr=Wcy3HHWN-=gwJKVqvdJnhjmS+px@mail.gmail.com>
	<AANLkTikuYdZbfoX3OX7UZTcq6A+kxcTpPU52gN8BT+vE@mail.gmail.com>
Message-ID: <AANLkTi=r1J8eztOrNQa4k5h1BFpVgm2uX2rsZAeESGnF@mail.gmail.com>

On Sat, Mar 19, 2011 at 3:22 PM, Rick DeNatale <rick.denatale at gmail.com> wrote:
> But this isn't doing the right thing in terms of setting the
> parameters in the rack request, you need to read and understand what's
> happening in the actionpack routing code in rails 3 to make this work.
>

By the way, part of the problem is that Rails 3 action_dispatch uses a
series of rack middlewares to set up the environment before you can
actually dispatch to a controller, by giving rails routing a lambda
instead of a string, all that gets bypassed so you need to do it
yourself.


-- 
Rick DeNatale

Blog: http://talklikeaduck.denhaven2.com/
Github: http://github.com/rubyredrick
Twitter: @RickDeNatale
WWR: http://www.workingwithrails.com/person/9021-rick-denatale
LinkedIn: http://www.linkedin.com/in/rickdenatale

From thomas at ravinggenius.com  Sat Mar 19 17:29:58 2011
From: thomas at ravinggenius.com (Thomas Ingram)
Date: Sat, 19 Mar 2011 17:29:58 -0400
Subject: [raleigh.rb] help with dynamically setting root
	controller#action
In-Reply-To: <AANLkTi=r1J8eztOrNQa4k5h1BFpVgm2uX2rsZAeESGnF@mail.gmail.com>
References: <AANLkTinSXYW2p5TGoCR+=iEWnQcmU4=rXR6jw+OX4QRP@mail.gmail.com>
	<2DC40691-B938-4BD8-87BE-1D1F4975AE9A@iannopollo.com>
	<AANLkTimmsi046ADr=Wcy3HHWN-=gwJKVqvdJnhjmS+px@mail.gmail.com>
	<AANLkTikuYdZbfoX3OX7UZTcq6A+kxcTpPU52gN8BT+vE@mail.gmail.com>
	<AANLkTi=r1J8eztOrNQa4k5h1BFpVgm2uX2rsZAeESGnF@mail.gmail.com>
Message-ID: <AANLkTi=NuRE0N92yp=ixjWxJE-cN3c5fmQrzyfzJs7cr@mail.gmail.com>

Thanks for the help guys; I got this working! I ended up going with a
controller so I wouldn't have to worry about setting up the
environment. Everything works as expected, but the way it is
implemented feels sub-optimal. I've updated the gist
(https://gist.github.com/877618) in case anyone wants to improve on
it. Specifically I'd like to the call to render. Oh well...

On Sat, Mar 19, 2011 at 15:26, Rick DeNatale <rick.denatale at gmail.com> wrote:
> By the way, part of the problem is that Rails 3 action_dispatch uses a
> series of rack middlewares to set up the environment before you can
> actually dispatch to a controller, by giving rails routing a lambda
> instead of a string, all that gets bypassed so you need to do it
> yourself.

-- 
Thomas Ingram ><>

From robbie at statsheet.com  Wed Mar 23 00:26:40 2011
From: robbie at statsheet.com (Robbie Allen)
Date: Wed, 23 Mar 2011 00:26:40 -0400
Subject: [raleigh.rb] Job Opportunity: Sports + Rails
Message-ID: <AANLkTin=n_AOd_AFOjeBpLjyrageRz_m1p=UJOcVFSvt@mail.gmail.com>

My company has a position open for a sports-loving, Rails developer with at
least 3 years experience with Ruby.  If you are interested in working at a
fast-paced, venture-backed startup in the Triangle that has one of the
largest Rails environments around, please send me an email:
robbie at statsheet.com

Regards,
Robbie Allen
Founder & CEO
http://statsheet.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110323/523d00a3/attachment.html>

From jeff at jumpstartlab.com  Wed Mar 23 16:44:56 2011
From: jeff at jumpstartlab.com (Jeff Casimir)
Date: Wed, 23 Mar 2011 16:44:56 -0400
Subject: [raleigh.rb] Ruby Jumpstart & Rails Jumpstart at the American
	Underground
Message-ID: <AANLkTi=vb3bxHnKiq=nhRK2cWfNAB0UfUvEu1pZUEv1B@mail.gmail.com>

Raleigh Rubyists,

Thanks for all your support and help recruiting attendees for my Ruby
and Rails Jumpstart classes.  I'm excited to be coming down and look
forward to scheduling more classes in your area (like intermediate
Rails, advanced Ruby, jQuery, etc).  It took WEEKS of emails and
calls, but I was able to relocate these classes to a more accessible
location at the American Underground in Durham
(http://www.americanunderground.com/).

Ruby Jumpstart, an introduction to Ruby for new programmers, is going
down this Saturday and Sunday:
http://jumpstartlab.com/trainings/14-ruby-durham

Rails Jumpstart, an introduction to Rails for programmers, follows it
on Monday & Tuesday: http://jumpstartlab.com/trainings/15-rails-durham

I'd particularly like to pickup one or two more people for the Ruby
class.  If you know someone who'd like to try out programming, please
put us in touch!  If the ticket price is a barrier I'm open to many
options like bartering a little work or whatever!  If you're already a
Ruby programmer and want to bring a friend, significant other, or kid
to the class and be their pair, you'd only pay for one ticket.  Plus
there's the discount code "raleigh" to knock 10% off any ticket.

Thanks,
Jeff

---
Jeff Casimir
Jumpstart Lab by Casimir Creative, LLC
http://jumpstartlab.com
@jumpstartlab on twitter

From sbauman at gmail.com  Wed Mar 23 17:19:27 2011
From: sbauman at gmail.com (Shannon Bauman)
Date: Wed, 23 Mar 2011 17:19:27 -0400
Subject: [raleigh.rb] Job opening at funded startup
Message-ID: <AANLkTinHHb-vw0ztVM3ruJnD+2wNgz1_OmSGAUm4Rx+H@mail.gmail.com>

Ruby Folks,

I've joined up with some other great folks in building a local startup
called Spring Metrics.  We now have an ex-Google engineer, an ex-Motricity
engineer, and an ex-Google Product Manager.  We are looking to bring on our
first full time RoR engineer to continue building out the strong engineering
team. We have a great location in downtown Durham, an unlimited vacation
policy, and our employees get stake in the early
funded<http://www.techjournalsouth.com/2011/03/durham-based-spring-metrics-nabs-635k-seed-round-to-spot-dna-of-conversions/>
 startup.

Full job posting is below as well as at:
http://www.springmetrics.com/jobs.html

If interested, send resumes to jobs at springmetrics.com.  Feel free to forward
on to others if you think they may be a good fit.

Thanks!
Shannon

======================


Ruby on Rails Engineer

Spring Metrics, a Durham-based Internet startup, is looking to hire our
first full-time Ruby on Rails engineer who will bring great ideas and
rock-solid execution to the front end of our well-regarded web analytics
service.

We are fun-loving brainiacs who live to solve difficult problems and enjoy
an informal setting where the best ideas carry the day. Our goal is to
assemble a team of creative dynamos who love to build products that are,
simply put, awesome.

If you are excited about the opportunity to shape the culture and products
of a company on the rise, drop us a line.

Responsibilities

   - Architect and code customer facing applications in Ruby on Rails
   - Own cutting edge JS code that will get used by millions of people
   - Brainstorm product features with product team

 Requirements

   - Fluent in Ruby on Rails (experience with Rails 3.x preferred)
   - Fluent in Javascript (non-jQuery)
   - Experience with jQuery
   - Significant experience with databases
   - Significant knowledge of HTML and CSS
   - Experience testing for the web (unit testing, browser compatibility
   testing)

 Nice-to-haves

   - Bachelors or Masters degree in Computer Science
   - Experience with version control systems (preferably Git)

 Still not sure if the position is right for you? Did we mention that our
office is in the heart of Durham?s thriving entrepreneurial scene? Or that
we?re within walking distance of tons of great restaurants? Or that we have
a foosball table and a hammock in our office? Oh, and that we happen to have
an unlimited vacation policy?

To submit an application, send a resume to jobs at springmetrics.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110323/ef22b095/attachment-0001.html>

From sbauman at gmail.com  Wed Mar 23 17:50:04 2011
From: sbauman at gmail.com (Shannon Bauman)
Date: Wed, 23 Mar 2011 17:50:04 -0400
Subject: [raleigh.rb] Job opening at funded startup
In-Reply-To: <AANLkTinHHb-vw0ztVM3ruJnD+2wNgz1_OmSGAUm4Rx+H@mail.gmail.com>
References: <AANLkTinHHb-vw0ztVM3ruJnD+2wNgz1_OmSGAUm4Rx+H@mail.gmail.com>
Message-ID: <AANLkTi=bFPmWQJAj3L5P_1Wf0qx_o4Wus+Zp_h5NFXne@mail.gmail.com>

P.S.

We're having an open house on Friday to celebrate our new office and our
round of funding.  If you think you might be interested in the job, or even
if you just want to swing by and say hi to the team - by all means come on
by.

211 N. Church St
Durham, NC

-shannon



On Wed, Mar 23, 2011 at 5:19 PM, Shannon Bauman <sbauman at gmail.com> wrote:

> Ruby Folks,
>
> I've joined up with some other great folks in building a local startup
> called Spring Metrics.  We now have an ex-Google engineer, an ex-Motricity
> engineer, and an ex-Google Product Manager.  We are looking to bring on our
> first full time RoR engineer to continue building out the strong engineering
> team. We have a great location in downtown Durham, an unlimited vacation
> policy, and our employees get stake in the early funded<http://www.techjournalsouth.com/2011/03/durham-based-spring-metrics-nabs-635k-seed-round-to-spot-dna-of-conversions/>
>  startup.
>
> Full job posting is below as well as at:
> http://www.springmetrics.com/jobs.html
>
> If interested, send resumes to jobs at springmetrics.com.  Feel free to
> forward on to others if you think they may be a good fit.
>
> Thanks!
> Shannon
>
> ======================
>
>
> Ruby on Rails Engineer
>
> Spring Metrics, a Durham-based Internet startup, is looking to hire our
> first full-time Ruby on Rails engineer who will bring great ideas and
> rock-solid execution to the front end of our well-regarded web analytics
> service.
>
> We are fun-loving brainiacs who live to solve difficult problems and enjoy
> an informal setting where the best ideas carry the day. Our goal is to
> assemble a team of creative dynamos who love to build products that are,
> simply put, awesome.
>
> If you are excited about the opportunity to shape the culture and products
> of a company on the rise, drop us a line.
>
> Responsibilities
>
>    - Architect and code customer facing applications in Ruby on Rails
>    - Own cutting edge JS code that will get used by millions of people
>    - Brainstorm product features with product team
>
>  Requirements
>
>    - Fluent in Ruby on Rails (experience with Rails 3.x preferred)
>    - Fluent in Javascript (non-jQuery)
>    - Experience with jQuery
>    - Significant experience with databases
>    - Significant knowledge of HTML and CSS
>    - Experience testing for the web (unit testing, browser compatibility
>    testing)
>
>  Nice-to-haves
>
>    - Bachelors or Masters degree in Computer Science
>    - Experience with version control systems (preferably Git)
>
>  Still not sure if the position is right for you? Did we mention that our
> office is in the heart of Durham?s thriving entrepreneurial scene? Or that
> we?re within walking distance of tons of great restaurants? Or that we have
> a foosball table and a hammock in our office? Oh, and that we happen to have
> an unlimited vacation policy?
>
> To submit an application, send a resume to jobs at springmetrics.com
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110323/015e9165/attachment.html>

From tom.brooke at gmail.com  Thu Mar 24 08:49:42 2011
From: tom.brooke at gmail.com (Tom Brooke)
Date: Thu, 24 Mar 2011 08:49:42 -0400
Subject: [raleigh.rb] Ruby Jumpstart & Rails Jumpstart at the American
	Underground
In-Reply-To: <AANLkTi=vb3bxHnKiq=nhRK2cWfNAB0UfUvEu1pZUEv1B@mail.gmail.com>
References: <AANLkTi=vb3bxHnKiq=nhRK2cWfNAB0UfUvEu1pZUEv1B@mail.gmail.com>
Message-ID: <AANLkTi=a0svrKpd6ospF8xUPr-fsgE41MJNeMV13dxbE@mail.gmail.com>

Jeff

I got a note from Jim Van Fleet about possible classes in Charlotte and
which would be nice and the Raleigh classes  but I guess in my case Raleigh
is about the same so I could go to either. I was in contact with you earlier
about Rails training in Raleigh and ended up going with Engine Yard which I
don't regret but yours might have been a better choice and I plan to do some
of yours in the future. The engine yard training was intense and it did get
me going on Rails but barely which I guess is all I can expect in 4 days.
The training had a heavy emphasis on rspec whic I know is a good thing but
as a beginner just playing with Rails I feel like I need to just get a feel
for rails and add the rspec later- They're philosophy was that even though
rspec was difficult at first I would be grateful in the long run. Also the
training was very fast whic again I guess was necessary since there was a
lot to cover

So as I continue I may be interested in advanced rails, advanced Ruby or
jquery so keep me posted as to Raleigh or Charlotte schedules

Keep me posted

Tom Brooke

On Wed, Mar 23, 2011 at 4:44 PM, Jeff Casimir <jeff at jumpstartlab.com> wrote:

> Raleigh Rubyists,
>
> Thanks for all your support and help recruiting attendees for my Ruby
> and Rails Jumpstart classes.  I'm excited to be coming down and look
> forward to scheduling more classes in your area (like intermediate
> Rails, advanced Ruby, jQuery, etc).  It took WEEKS of emails and
> calls, but I was able to relocate these classes to a more accessible
> location at the American Underground in Durham
> (http://www.americanunderground.com/).
>
> Ruby Jumpstart, an introduction to Ruby for new programmers, is going
> down this Saturday and Sunday:
> http://jumpstartlab.com/trainings/14-ruby-durham
>
> Rails Jumpstart, an introduction to Rails for programmers, follows it
> on Monday & Tuesday: http://jumpstartlab.com/trainings/15-rails-durham
>
> I'd particularly like to pickup one or two more people for the Ruby
> class.  If you know someone who'd like to try out programming, please
> put us in touch!  If the ticket price is a barrier I'm open to many
> options like bartering a little work or whatever!  If you're already a
> Ruby programmer and want to bring a friend, significant other, or kid
> to the class and be their pair, you'd only pay for one ticket.  Plus
> there's the discount code "raleigh" to knock 10% off any ticket.
>
> Thanks,
> Jeff
>
> ---
> Jeff Casimir
> Jumpstart Lab by Casimir Creative, LLC
> http://jumpstartlab.com
> @jumpstartlab on twitter
> _______________________________________________
> raleigh-rb-members mailing list
> raleigh-rb-members at rubyforge.org
> http://rubyforge.org/mailman/listinfo/raleigh-rb-members
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110324/e0c9729e/attachment-0001.html>

From mikekeung at gmail.com  Thu Mar 24 19:54:55 2011
From: mikekeung at gmail.com (Michael Keung)
Date: Thu, 24 Mar 2011 19:54:55 -0400
Subject: [raleigh.rb] Startup looking for developer / tech strategist
Message-ID: <000001cbea7e$e1712a60$a4537f20$@com>

Hi all,

 

I am currently developing an online fitness platform and am looking for
developers who are interested in helping realize the vision.  The platform
focuses on developing interconnected applications and incorporating social
networking aspects.

 

The role would be to develop the concepts using Ruby on Rails but
individuals are expected to play a major role in defining the company
strategy and concept implementation.  We want developers who are
enthusiastic in the computer science field or enthusiastic about fitness.
Developers would gain equity in the company.

 

If interested, please contact me at mikekeung at gmail.com with a portfolio of
your work.  

 

Best regards,


Michael Keung

Co-founder 
Evitca.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/raleigh-rb-members/attachments/20110324/b8063ed8/attachment.html>