From: Erlend Oye <segfault.ruby-talk@...>
Date: 2005-01-13T15:34:36+09:00
Subject: Re: Segfault in timer.rb

Erlend Oye wrote:

> Hi Florian,
>
> > Florian Gross wrote:
> >> Erlend Oye wrote:
> >> /usr/local/lib/ruby/1.8/timeout.rb:40: [BUG] rb_gc_mark() called for
> >> broken object
> >> ruby 1.8.2 (2005-01-10) [i386-cygwin]
>
> >> [...]
> >> zsh: 22780 abort (core dumped) ruby scrape_server.rb
> >> --------
> >> /usr/local/lib/ruby/1.8/timeout.rb:40: [BUG] rb_gc_mark(): unknown 
> data
> >> type 0x2e(0x1037d178) non object
> >> ruby 1.8.2 (2005-01-10) [i386-cygwin]
> >
>
> >Any other libraries that it's using? YAML used to do something similar.
>
> Here are the various requires and includes (not in order)
>
> require 'cgi'
> require 'net/http'
> require 'rexml/document'
> require 'uri/http'
> include WEBrick
> require 'webrick'
>
> I believe timeout is being used by Webrick internally and not RSSScraper
> directly.
>
> RSSScraper is at
>
> http://rssscraper.rubyforge.org
>
> Thanks
> Erlend Oye

All of these segfaults seemed to be happening at

Thread.start

statements. This hinted some trouble with thread support so I recompiled 
using

--disable-pthread

and the troubles have disappeared. I wonder if this is an issue from 
Ruby's side
or on Cygwin's side. I am noticing other segfaults being  reported on 
other platforms
(e.g linux) due to pthreads support also.

Thanks
Erlend Oye