Ruby 1.8.7 p330 ERRNO out of memory errors

[Drew]

I'm posting this here as an expansion to what I posted at this link (at Luis' request) http://www.rubyinside.com/rails-installer-ruby-and-rails-on-windows-in-a-single-install-4201.html

I suspect that this is not a RubyInstaller issue though I concede it is possible. A few weeks ago I got a new computer at work (moved from win vista 32bit to win 7 64bit) and around the same time Ruby 1.8.7 p330 came out so I decided to install that. Our in-house rails project has some views that are....well let's just say poorly optimized. They give lots of output to the console where lots is hundreds of lines of output (I'm not sure on exact numbers). In any case this bug is really easy to replicate and I even found a related issue on the Ruby issue tracker here (http://redmine.ruby-lang.org/issues/show/1063) except that I am not on XP and this isn't 1.8.6 or 1.9.1.

I use pik to manage my rubies so here's a code sample - this also demonstrates that this is not a rails issue:

C:\Users\drew>pik 187

Select which Ruby you want:

1. 187: ruby 1.8.7 (2010-08-16 patchlevel 302) [i386-mingw32]

2. 187: ruby 1.8.7 (2010-12-23 patchlevel 330) [i386-mingw32]

?  2

C:\Users\drew>ruby -v

ruby 1.8.7 (2010-12-23 patchlevel 330) [i386-mingw32]

C:\Users\drew>irb

> STDOUT.write "a"*100000000

Errno::ENOMEM: Not enough space

        from (irb):1:in `write'

        from (irb):1

This exact same line of code works fine on 1.8.7 p302, the previous stable for 187. As a result of this, I cannot use p330 with our in-house rails projects because requests just die. Any insight is greatly appreciated!

[Luis Lavena]

On Tue, Jan 18, 2011 at 1:18 PM, Drew <lato...@gmail.com> wrote:
> I'm posting this here as an expansion to what I posted at this link (at
> Luis'
> request) http://www.rubyinside.com/rails-installer-ruby-and-rails-on-windows-in-a-single-install-4201.html

Thank you Drew.

> In any case this bug is really easy to replicate and I even found
> a related issue on the Ruby issue tracker here
> (http://redmine.ruby-lang.org/issues/show/1063) except that I am not on XP
> and this isn't 1.8.6 or 1.9.1.

1.8.7 inherits a lot of 1.8.6 code (both share the same codebase)
1.9.1 had the same issue for console devices (STDOUT and STDERR)

If you mention this didn't happen on previous version, then is a regression.

>
> This exact same line of code works fine on 1.8.7 p302, the previous stable
> for 187. As a result of this, I cannot use p330 with our in-house rails
> projects because requests just die. Any insight is greatly appreciated!
>

Understood. I'll try to replicate in all the available versions and
report back to Ruby-Core.

Thank you for your time posting this, is really helpful.

--
Luis Lavena
AREA 17
-
Perfection in design is achieved not when there is nothing more to add,
but rather when there is nothing more to take away.
Antoine de Saint-Exupéry

[Luis Lavena]

On Tue, Jan 18, 2011 at 1:34 PM, Luis Lavena <luisl...@gmail.com> wrote:
>
> If you mention this didn't happen on previous version, then is a regression.
>

> Understood. I'll try to replicate in all the available versions and
> report back to Ruby-Core.
>

I've reported the regression:

http://redmine.ruby-lang.org/issues/show/4293

And also indicated where in the code the break was introduced.

Assigned to Nobu for confirmation and possible thoughts about the fix.

Will keep you posted.

[Drew]

Excellent, thank you very much for your work on this!

[Luis Lavena]

On Wed, Jan 19, 2011 at 2:14 PM, Drew <lato...@gmail.com> wrote:
> Excellent, thank you very much for your work on this!

A real pleasure.

FYI, seems the issue got solved and a fix will be backported into
1.8.7 pretty soon.

I hope in a month or so a new patchlevel will go out.

[rogerdpack]

> FYI, seems the issue got solved and a fix will be backported into
> 1.8.7 pretty soon.

So the issue was solved but the redmine ticket is still set to
"assigned" is that right (somebody just needs to close it)?

http://redmine.ruby-lang.org/issues/show/4293

Thanks!
-r

[Luis Lavena]

Is pending to be backported to ruby_1_8_7 branch. As you can see is
assigned to Shyouhei Urabe, maintainer of the 1.8.7 release.

I've just updated the ticket with a ping to Urabe-san.

I don't want to step into other's roles at Core.