From thewoolleyman at gmail.com  Mon Oct  5 02:16:10 2009
From: thewoolleyman at gmail.com (Chad Woolley)
Date: Sun, 4 Oct 2009 23:16:10 -0700
Subject: [Cruisecontrolrb-developers] Help with Git hanging bugs
In-Reply-To: <e3af97510907281835k23354badka747a7f24a1951a8@mail.gmail.com>
References: <e3af97510907281835k23354badka747a7f24a1951a8@mail.gmail.com>
Message-ID: <e3af97510910042316hc06ef7cue72af8a3f0a9b2db@mail.gmail.com>

On Tue, Jul 28, 2009 at 6:35 PM, Chad Woolley <thewoolleyman at gmail.com> wrote:
> We've taken a shot at fixing this with a timeout loop, but it still
> happens sometimes. ? ?It only happens in certain scenarios, and is
> hard (impossible for me so far) to reproduce. ?I've recorded the info
> I have in the comments of the ticket. ?Any help or further info is
> welcome:
>
> https://cruisecontrolrb.lighthouseapp.com/projects/9150/tickets/229-sometimes-git-hangs

Matt Duncan has helped track this down, but the fix will involve the
use of the SystemTimer native gem.  I plan to use Bundler to manage
this.  Please see the ticket if you'd like to discuss:

https://cruisecontrolrb.lighthouseapp.com/projects/9150-cruise-control-rb/tickets/229-sometimes-git-hangs#ticket-229-14

-- Chad

From thewoolleyman at gmail.com  Mon Oct  5 12:53:09 2009
From: thewoolleyman at gmail.com (Chad Woolley)
Date: Mon, 5 Oct 2009 09:53:09 -0700
Subject: [Cruisecontrolrb-developers] Help with Git hanging bugs
In-Reply-To: <3945c4270910050903j5290f886p95dfa6760ce0c1d4@mail.gmail.com>
References: <e3af97510907281835k23354badka747a7f24a1951a8@mail.gmail.com> 
	<e3af97510910042316hc06ef7cue72af8a3f0a9b2db@mail.gmail.com> 
	<3945c4270910050903j5290f886p95dfa6760ce0c1d4@mail.gmail.com>
Message-ID: <e3af97510910050953i372a8173h2269cb83bbf4ed40@mail.gmail.com>

On Mon, Oct 5, 2009 at 9:03 AM, Alexey Verkhovsky
<alexey.verkhovsky at gmail.com> wrote:
> I don't like the sound of automaticaly installing native gems. This screams
> "maintenance nightmare" to me (think Windoze). Why don't we leave
> "SystemTimer not installed" as a fairly visible warning (written to the
> dashboard, and linked to a documentation page that explains the whole thing,
> perhaps)? This way, if a build gets stuck, people will easily find out why
> and take corrective action.

We definitely can and should do that, but I'd still like to have a
fail-safe installation attempt, just for my own selfish automation
purposes.

From thewoolleyman at gmail.com  Mon Oct  5 13:17:48 2009
From: thewoolleyman at gmail.com (Chad Woolley)
Date: Mon, 5 Oct 2009 10:17:48 -0700
Subject: [Cruisecontrolrb-developers] Help with Git hanging bugs
In-Reply-To: <3945c4270910051008w2f1cfb05kc998f2f490c449c5@mail.gmail.com>
References: <e3af97510907281835k23354badka747a7f24a1951a8@mail.gmail.com> 
	<e3af97510910042316hc06ef7cue72af8a3f0a9b2db@mail.gmail.com> 
	<3945c4270910050903j5290f886p95dfa6760ce0c1d4@mail.gmail.com> 
	<e3af97510910050953i372a8173h2269cb83bbf4ed40@mail.gmail.com> 
	<3945c4270910051008w2f1cfb05kc998f2f490c449c5@mail.gmail.com>
Message-ID: <e3af97510910051017y59417b6au30ef787ff9909ec0@mail.gmail.com>

On Mon, Oct 5, 2009 at 10:08 AM, Alexey Verkhovsky
<alexey.verkhovsky at gmail.com> wrote:
> (1) computers behind paranoid firewalls (by adding network timeout to every
> attempt to start an instance, which may or may not be infinite).
> (2) computers that don't have gcc or ruby-dev (increasing startup time)
> (3) you'll need to address what happens when several CC processes start from
> the same install at the same time (because, say, I like to run builder
> processes separately from the dashboard).
> (4) possibly, computers that run CC.rb under nologin accounts.
> (5) for people who don't want the system_timer gem for any reason
> whatsoever.
> (6) ???
>
> Seriously, a shrink-wrapped product should not silently install stuff into
> any global library locations, IMHO.

Good points.  One note though - with Bundler, it doesn't install to a
global location, it is installed under vendor/gems.  We can also do a
platform check, plus there is a 'placebo' stub gem on Windows.

Maybe we can make the bundling automated, but only if there is an
"opt-in" flag in cruise config?  Or maybe do it on all supported
platforms unless there is an "opt-out" flag set?

From thewoolleyman at gmail.com  Mon Oct  5 13:20:41 2009
From: thewoolleyman at gmail.com (Chad Woolley)
Date: Mon, 5 Oct 2009 10:20:41 -0700
Subject: [Cruisecontrolrb-developers] (no subject)
In-Reply-To: <3945c4270910050959y5dcf4b82oe22c672c279de4c2@mail.gmail.com>
References: <3945c4270910050959y5dcf4b82oe22c672c279de4c2@mail.gmail.com>
Message-ID: <e3af97510910051020h38dff5d2jfde3afff0f4d9db8@mail.gmail.com>

On Mon, Oct 5, 2009 at 9:59 AM, Alexey Verkhovsky
<alexey.verkhovsky at gmail.com> wrote:
> http://betarelease.github.com/2009/10/04/cruisecontrol-monitors-gotchas.html
>
> We should totally
>
> * make a saner alternative URL for the build status XML document (the .aspx
> thing harks back to when the only available dashboard monitor option was
> from CruiseControl.NET world).
> * display a rather prominent link to the aforesaid document from the /builds
> page. Probably also add an XML comment to it, explaining what it is.

Make a ticket in Lighthouse!

BTW, I think LightHouse kinda sucks.  People have asked them for
months to allow emails notifications (such as this list) on new ticket
creation, and they still have not added it.  Seems like they are just
letting it stagnate and not doing many new features.

From ckponnappa at gmail.com  Wed Oct 28 22:56:22 2009
From: ckponnappa at gmail.com (C. K. Ponnappa)
Date: Thu, 29 Oct 2009 08:26:22 +0530
Subject: [Cruisecontrolrb-developers] [GitHub] valo sent you a message
In-Reply-To: <4ae8b72d2e1e7_17ed3fe199347af04fc@fe4.rs.github.com.tmail>
References: <4ae8b72d2e1e7_17ed3fe199347af04fc@fe4.rs.github.com.tmail>
Message-ID: <4AE90456.3040005@gmail.com>

Hey Valentin,

Your patch looks good, but I have 7 failing tests after pulling; I just 
wanted to confirm if this is happening for you (looks like a missing 
cattr_reader :output_limit in Configuration)?
We can merge it in once the others have approved the change.

Thanks,
Sidu.
http://blog.sidu.in
http://twitter.com/ponnappa

GitHub wrote:
> valo wants you to pull from valo/cruisecontrol.rb at master
>
> Body: Hi there!
>
> I made a patch to make the output limit of 100KB configurable. It is a global config option on the Configuration class. If you want and like the patch you can pull it.
>
> Please tell me if the fix is not good enough.
>
> Thanks,
>   Valentin
>
> View repository: http://github.com/valo/cruisecontrol.rb/tree/master
>
>   

From ckponnappa at gmail.com  Thu Oct 29 16:11:04 2009
From: ckponnappa at gmail.com (C. K. Ponnappa)
Date: Fri, 30 Oct 2009 01:41:04 +0530
Subject: [Cruisecontrolrb-developers] Making output limit configurable
Message-ID: <4AE9F6D8.5050104@gmail.com>

Folks,

Shall I go ahead and merge Valentin's patch to make the output size 
configurable? You can find his changes at  
http://github.com/valo/cruisecontrol.rb

Thanks,
Sidu.

-------- Original Message --------
Subject: 	Re: [GitHub] valo sent you a message
Date: 	Fri, 30 Oct 2009 01:37:50 +0530
From: 	C. K. Ponnappa <ckponnappa at gmail.com>
To: 	Valentin Mihov <valentin.mihov at gmail.com>
References: 	<4ae8b72d2e1e7_17ed3fe199347af04fc at fe4.rs.github.com.tmail> 
<4AE90456.3040005 at gmail.com> 
<45448450910290252h17a1c018ife55a1ffa194ca1b at mail.gmail.com>



Sure Valentin, more tests are always welcome. In the meanwhile I will 
get Alexey and others to review and once approved, I will merge changes.

Thanks ever so much!
Sidu.
http://blog.sidu.in
http://twitter.com/ponnappa

Valentin Mihov wrote:
> OK. This time the unit tests are not failing. I have forgotten to
> declare the assessors for the output_limit. I changed it in our CI,
> but forgot to push it to github. It should be better now.
>
> Should I write some unit tests for the new config var?
>
> Here is the fix:
> http://github.com/valo/cruisecontrol.rb/commit/e1037bff7278ba1471dd4b91b88ee8d86bb19f14
>
> Thanks,
>   Valentin
>
> On Thu, Oct 29, 2009 at 4:56 AM, C. K. Ponnappa <ckponnappa at gmail.com> wrote:
>   
>> Hey Valentin,
>>
>> Your patch looks good, but I have 7 failing tests after pulling; I just
>> wanted to confirm if this is happening for you (looks like a missing
>> cattr_reader :output_limit in Configuration)?
>> We can merge it in once the others have approved the change.
>>
>> Thanks,
>> Sidu.
>> http://blog.sidu.in
>> http://twitter.com/ponnappa
>>
>> GitHub wrote:
>>     
>>> valo wants you to pull from valo/cruisecontrol.rb at master
>>>
>>> Body: Hi there!
>>>
>>> I made a patch to make the output limit of 100KB configurable. It is a
>>> global config option on the Configuration class. If you want and like the
>>> patch you can pull it.
>>>
>>> Please tell me if the fix is not good enough.
>>>
>>> Thanks,
>>>  Valentin
>>>
>>> View repository: http://github.com/valo/cruisecontrol.rb/tree/master
>>>
>>>
>>>       
>
>   



From thewoolleyman at gmail.com  Thu Oct 29 18:52:43 2009
From: thewoolleyman at gmail.com (Chad Woolley)
Date: Thu, 29 Oct 2009 15:52:43 -0700
Subject: [Cruisecontrolrb-developers] Making output limit configurable
In-Reply-To: <4AE9F6D8.5050104@gmail.com>
References: <4AE9F6D8.5050104@gmail.com>
Message-ID: <e3af97510910291552u3e3e46cbp85b213898aba6c5c@mail.gmail.com>

On Thu, Oct 29, 2009 at 1:11 PM, C. K. Ponnappa <ckponnappa at gmail.com> wrote:
> Folks,
>
> Shall I go ahead and merge Valentin's patch to make the output size
> configurable? You can find his changes at
> ?http://github.com/valo/cruisecontrol.rb

I'd feel better about it if the build were green and building reliably:

http://cruisecontrolrb.thoughtworks.com/builds/CruiseControlRB

-- Chad