From tbray at textuality.com  Thu Jul  3 18:47:00 2008
From: tbray at textuality.com (Tim Bray)
Date: Thu, 3 Jul 2008 15:47:00 -0700
Subject: [Ape-development] Testing
Message-ID: <517bf110807031547o450b59eara9b8150eb27253f1@mail.gmail.com>

Is this thing on?  -T

From tbray at textuality.com  Thu Jul  3 19:22:29 2008
From: tbray at textuality.com (Tim Bray)
Date: Thu, 3 Jul 2008 16:22:29 -0700
Subject: [Ape-development] Recent Ape not re-entrant?
Message-ID: <517bf110807031622u21318073mef59a5c738a2dfd5@mail.gmail.com>

I'm using Ape to debug a server implementation... works fine first
time I run it, crashes 2nd time:
=======================
The Ape says:

http://localhost:4444/app/atom/service

Summary: 1 error, 0 warnings

   1. TESTING: Service document and collections.
   2. ? Retrieval of Service Document: it exists and is served properly.[Dialog]
   3. Schema validation is just available building the ape with jruby.
   4.

      Found these collections:
          * 'Entries' accepts application/atom+xml;type=entry
          * 'Media' accepts */*
   5. ! Ouch! Ape fall down go boom; details: haven't been setted
      Ape::ValidationError
      ./bin/../lib/ape.rb:147:in
`check_manifest'./bin/../lib/ape.rb:159:in
`check_validator'./bin/../lib/ape.rb:75:in
`might_fail'./bin/../lib/ape.rb:74:in `each'./bin/../lib/ape.rb:74:in
`might_fail'./bin/../lib/ape.rb:59:in
`check'./bin/../lib/ape/handler.rb:31:in
`process'/System/Library/Frameworks/Ruby.framework/Versions/1.8/usr/lib/ruby/gems/1.8/gems/mongrel-1.1.4/lib/mongrel.rb:159:in
==========================
I'm poking around trying to figure out this manifest thing, thought
I'd post this while I do that in case someone says "Oh, that's
obvious"
 -Tim

From david.calavera at gmail.com  Fri Jul  4 03:51:20 2008
From: david.calavera at gmail.com (David Calavera)
Date: Fri, 4 Jul 2008 09:51:20 +0200
Subject: [Ape-development] Recent Ape not re-entrant?
In-Reply-To: <517bf110807031622u21318073mef59a5c738a2dfd5@mail.gmail.com>
References: <517bf110807031622u21318073mef59a5c738a2dfd5@mail.gmail.com>
Message-ID: <baf7e05c0807040051n467ce058jf53d3dadd29d8107@mail.gmail.com>

It seem that the check_manifest method deleted the class fields moreover the
variable field. I've fixed and I've released a new version of the gem
because I think is a critical bug.

On Fri, Jul 4, 2008 at 1:22 AM, Tim Bray <tbray at textuality.com> wrote:

> I'm using Ape to debug a server implementation... works fine first
> time I run it, crashes 2nd time:
> =======================
> The Ape says:
>
> http://localhost:4444/app/atom/service
>
> Summary: 1 error, 0 warnings
>
>   1. TESTING: Service document and collections.
>   2. ? Retrieval of Service Document: it exists and is served
> properly.[Dialog]
>   3. Schema validation is just available building the ape with jruby.
>   4.
>
>      Found these collections:
>          * 'Entries' accepts application/atom+xml;type=entry
>          * 'Media' accepts */*
>   5. ! Ouch! Ape fall down go boom; details: haven't been setted
>      Ape::ValidationError
>      ./bin/../lib/ape.rb:147:in
> `check_manifest'./bin/../lib/ape.rb:159:in
> `check_validator'./bin/../lib/ape.rb:75:in
> `might_fail'./bin/../lib/ape.rb:74:in `each'./bin/../lib/ape.rb:74:in
> `might_fail'./bin/../lib/ape.rb:59:in
> `check'./bin/../lib/ape/handler.rb:31:in
>
> `process'/System/Library/Frameworks/Ruby.framework/Versions/1.8/usr/lib/ruby/gems/1.8/gems/mongrel-1.1.4/lib/mongrel.rb:159:in
> ==========================
> I'm poking around trying to figure out this manifest thing, thought
> I'd post this while I do that in case someone says "Oh, that's
> obvious"
>  -Tim
> _______________________________________________
> Ape-development mailing list
> Ape-development at rubyforge.org
> http://rubyforge.org/mailman/listinfo/ape-development
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/ape-development/attachments/20080704/9dfe7c5d/attachment.html>

From tbray at textuality.com  Fri Jul 11 09:59:08 2008
From: tbray at textuality.com (Tim Bray)
Date: Fri, 11 Jul 2008 06:59:08 -0700
Subject: [Ape-development] Thinking about how to use Ape
Message-ID: <517bf110807110659h3aa0667s8dc8b64b07c74d33@mail.gmail.com>

I've got meta-publication (a collection that you post to do collection
CRUD) basically working in mod_atom, and I've been using the Ape, but
it's kind of clumsy, and with the new metaprogramming hooks (invokers
and validators and so on) I'm wondering what hooks might make all this
more useful.   This is a request for ideas.

Here's how it works.  A meta-collection ("meta" for short) has a
special attribute mod-atom:meta="service" on its <atom:feed> root.  If
you post an atom:entry to it, it creates a couple of collections for
entries and media and a service doc to wrap them, let's call that a
sub-pub or sub for short.  So what I do with the Ape is:

- manually run it against the meta-pub.
- peek in the logs to see where one of the the newly-created pub went
(since the Ape insists on cleaning up its creations)
- manually run the ape against the newly-created Pub (and expect
everything to pass).

Now, the next thing I'm going to do when I finish cleaning up this
code is run some benchmarks.

So I think what I really want is a nice programmatic way to access the
Ape code to look inside service docs and collection docs, and then to
issue POSTs and PUTs and GETs and DELETEs.  Basically, find the
collection with the mod-atom:meta attribute, run a POST against it to
create a new pub, then, run Ape#check against the newly-created
service doc, then DELETE the new publication.  Then for benchmarking,
I want to create a huge number of new pubs and run a huge number of
random CRUD operations against them.

So if the Ape class gets a few more methods for this stuff, I just
need to have an Ape object for the meta and one for each of the subs
and with access to the CRUD methods, it should be pretty easy.

Now that I've written this, I've convinced myself that it shouldn't be
complicated, so I'll go start looking at how to do it.

I suspect that a general-purpose programmable Atompub client will be
quite a useful thing, and Ape is most of the way there now.  -Tim

From david.calavera at gmail.com  Fri Jul 11 17:26:39 2008
From: david.calavera at gmail.com (David Calavera)
Date: Fri, 11 Jul 2008 23:26:39 +0200
Subject: [Ape-development] Thinking about how to use Ape
In-Reply-To: <517bf110807110659h3aa0667s8dc8b64b07c74d33@mail.gmail.com>
References: <517bf110807110659h3aa0667s8dc8b64b07c74d33@mail.gmail.com>
Message-ID: <baf7e05c0807111426r3e615888s42a83e4fb77f334d@mail.gmail.com>

Well, I've followed the thread into the atom list XD.

Although now the ape is closer a testing tool instead of a library, you can
also use the core classes. Nevetheless I suppose we should improve and
publish the rdoc.

Btw, I send you a piece of code that I think it could be help you:

http://pastie.org/232348

On Fri, Jul 11, 2008 at 3:59 PM, Tim Bray <tbray at textuality.com> wrote:

> I've got meta-publication (a collection that you post to do collection
> CRUD) basically working in mod_atom, and I've been using the Ape, but
> it's kind of clumsy, and with the new metaprogramming hooks (invokers
> and validators and so on) I'm wondering what hooks might make all this
> more useful.   This is a request for ideas.
>
> Here's how it works.  A meta-collection ("meta" for short) has a
> special attribute mod-atom:meta="service" on its <atom:feed> root.  If
> you post an atom:entry to it, it creates a couple of collections for
> entries and media and a service doc to wrap them, let's call that a
> sub-pub or sub for short.  So what I do with the Ape is:
>
> - manually run it against the meta-pub.
> - peek in the logs to see where one of the the newly-created pub went
> (since the Ape insists on cleaning up its creations)
> - manually run the ape against the newly-created Pub (and expect
> everything to pass).
>
> Now, the next thing I'm going to do when I finish cleaning up this
> code is run some benchmarks.
>
> So I think what I really want is a nice programmatic way to access the
> Ape code to look inside service docs and collection docs, and then to
> issue POSTs and PUTs and GETs and DELETEs.  Basically, find the
> collection with the mod-atom:meta attribute, run a POST against it to
> create a new pub, then, run Ape#check against the newly-created
> service doc, then DELETE the new publication.  Then for benchmarking,
> I want to create a huge number of new pubs and run a huge number of
> random CRUD operations against them.
>
> So if the Ape class gets a few more methods for this stuff, I just
> need to have an Ape object for the meta and one for each of the subs
> and with access to the CRUD methods, it should be pretty easy.
>
> Now that I've written this, I've convinced myself that it shouldn't be
> complicated, so I'll go start looking at how to do it.
>
> I suspect that a general-purpose programmable Atompub client will be
> quite a useful thing, and Ape is most of the way there now.  -Tim
> _______________________________________________
> Ape-development mailing list
> Ape-development at rubyforge.org
> http://rubyforge.org/mailman/listinfo/ape-development
>



-- 
David Calavera
http://www.thinkincode.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/ape-development/attachments/20080711/103b48a4/attachment.html>

From tbray at textuality.com  Mon Jul 14 19:37:55 2008
From: tbray at textuality.com (Tim Bray)
Date: Mon, 14 Jul 2008 16:37:55 -0700
Subject: [Ape-development] Thinking about how to use Ape
In-Reply-To: <baf7e05c0807111426r3e615888s42a83e4fb77f334d@mail.gmail.com>
References: <517bf110807110659h3aa0667s8dc8b64b07c74d33@mail.gmail.com>
	<baf7e05c0807111426r3e615888s42a83e4fb77f334d@mail.gmail.com>
Message-ID: <517bf110807141637u5581b0b6h215f692bbeadd270@mail.gmail.com>

2008/7/11 David Calavera <david.calavera at gmail.com>:

> Although now the ape is closer a testing tool instead of a library, you can
> also use the core classes. Nevetheless I suppose we should improve and
> publish the rdoc.
>
> Btw, I send you a piece of code that I think it could be help you:
>
> http://pastie.org/232348

Here's a little pastie with my thinking... I don't think something
like this would be too hard to build.  I think that the Ape could
quite easily become the Ruby standard Atompub client library, not just
the testing library.

http://pastie.org/233389

 -Tim

From tbray at textuality.com  Tue Jul 15 10:52:10 2008
From: tbray at textuality.com (Tim Bray)
Date: Tue, 15 Jul 2008 07:52:10 -0700
Subject: [Ape-development] libxml
Message-ID: <517bf110807150752j4a2a7ff1m8cd2de4fc9b772b5@mail.gmail.com>

I hate REXML more than you, but please don't change to libxml... it
won't work on JRuby.

Thanks, Tim

From david.calavera at gmail.com  Tue Jul 15 11:02:50 2008
From: david.calavera at gmail.com (David Calavera)
Date: Tue, 15 Jul 2008 17:02:50 +0200
Subject: [Ape-development] performance REXML vs libxml-ruby
Message-ID: <baf7e05c0807150802h19b4a8f4of348ab25c373aed7@mail.gmail.com>

Hi,

I've created a branch where I migrated the xml code to libxml-ruby. I've ran
a benchmark test against the Joe Gregorio apptestsite. I endorse the
results:

ape-trunk$ ruby benchmark.rb '
http://bitworking.org/projects/apptestsite/app.cgi/service/;service_document
'
      user     system      total        real
  1.950000   0.220000   2.170000 ( 73.387951)

ape-libxml$ ruby benchmark.rb '
http://bitworking.org/projects/apptestsite/app.cgi/service/;service_document
'
      user     system      total        real
  0.570000   0.090000   0.660000 ( 88.123802)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/ape-development/attachments/20080715/bd84bc53/attachment.html>

From david.calavera at gmail.com  Tue Jul 15 16:36:47 2008
From: david.calavera at gmail.com (David Calavera)
Date: Tue, 15 Jul 2008 22:36:47 +0200
Subject: [Ape-development] Thinking about how to use Ape
In-Reply-To: <517bf110807141637u5581b0b6h215f692bbeadd270@mail.gmail.com>
References: <517bf110807110659h3aa0667s8dc8b64b07c74d33@mail.gmail.com>
	<baf7e05c0807111426r3e615888s42a83e4fb77f334d@mail.gmail.com>
	<517bf110807141637u5581b0b6h215f692bbeadd270@mail.gmail.com>
Message-ID: <baf7e05c0807151336g652489b4qdc6e4cb5677cb3f5@mail.gmail.com>

I'd say that methods can be really useful although I find the lazy loading a
little weird and I absolutly agree that we don't need the ape class.

On Tue, Jul 15, 2008 at 1:37 AM, Tim Bray <tbray at textuality.com> wrote:

> 2008/7/11 David Calavera <david.calavera at gmail.com>:
>
> > Although now the ape is closer a testing tool instead of a library, you
> can
> > also use the core classes. Nevetheless I suppose we should improve and
> > publish the rdoc.
> >
> > Btw, I send you a piece of code that I think it could be help you:
> >
> > http://pastie.org/232348
>
> Here's a little pastie with my thinking... I don't think something
> like this would be too hard to build.  I think that the Ape could
> quite easily become the Ruby standard Atompub client library, not just
> the testing library.
>
> http://pastie.org/233389
>
>  -Tim
>



-- 
David Calavera
http://www.thinkincode.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rubyforge.org/pipermail/ape-development/attachments/20080715/7aa01829/attachment.html>

From tbray at textuality.com  Sat Jul 19 12:38:17 2008
From: tbray at textuality.com (Tim Bray)
Date: Sat, 19 Jul 2008 09:38:17 -0700
Subject: [Ape-development] What does rake do?
Message-ID: <517bf110807190938x77d2887ft95f4b898f68b4f41@mail.gmail.com>

If I type "rake test" I get a messy blow-up.  There are a lot of
testfiles.  What's the setup?  -Tim

From david.calavera at gmail.com  Sat Jul 19 14:16:15 2008
From: david.calavera at gmail.com (David Calavera)
Date: Sat, 19 Jul 2008 20:16:15 +0200
Subject: [Ape-development] What does rake do?
In-Reply-To: <517bf110807190938x77d2887ft95f4b898f68b4f41@mail.gmail.com>
References: <517bf110807190938x77d2887ft95f4b898f68b4f41@mail.gmail.com>
Message-ID: <88E501AC-480E-4460-8A55-773C1F0607E3@gmail.com>

There isn't a setup file. It executes all the test files under test/ 
unit. You can look at Rakefile.

Enviado desde mi iPhone

El 19/07/2008, a las 18:38, "Tim Bray" <tbray at textuality.com> escribi?:

> If I type "rake test" I get a messy blow-up.  There are a lot of
> testfiles.  What's the setup?  -Tim
> _______________________________________________
> Ape-development mailing list
> Ape-development at rubyforge.org
> http://rubyforge.org/mailman/listinfo/ape-development

From tbray at textuality.com  Sun Jul 20 17:40:27 2008
From: tbray at textuality.com (Tim Bray)
Date: Sun, 20 Jul 2008 14:40:27 -0700
Subject: [Ape-development] Collections
Message-ID: <517bf110807201440w31f1893bvc4fc2aee0f60cee0@mail.gmail.com>

I think I realized why I kept re-re-re-re-factoring.  There are two
entirely different things, here.  The first is the <app:collection>
element in the service doc, the second is the thing that you send HTTP
messages to do resource CRUD.  The first should have an accept?
method, the second post/put/delete methods.  Ape::Collection should be
the second, since that's at the core of the protocol, with feed-paging
methods as well as the CRUD.

It's probably worth having a separate class, call it ServiceColl or
CollElement, for the second, because that's a good place to put
accept? and also (later) accept_cat? - so accept? becomes accept_type?

Sigh, blowing up lots of unit tests :(  -T

From tbray at textuality.com  Mon Jul 21 18:55:56 2008
From: tbray at textuality.com (Tim Bray)
Date: Mon, 21 Jul 2008 15:55:56 -0700
Subject: [Ape-development] ape_server in debug mode?
Message-ID: <517bf110807211555x64b84319q73a24c71089e1b98@mail.gmail.com>

I'm a mongrel n00b.  Is there a way to run ape_server in debug mode so
it'll reload changed .rb files?  -T