Subject: Re: pkg/34859: ruby with native pthread broken
To: None <taca@NetBSD.org, gnats-admin@netbsd.org, pkgsrc-bugs@netbsd.org,>
From: Takayoshi Kochi <takayoshi.kochi@gmail.com>
List: pkgsrc-bugs
Date: 10/20/2006 09:45:02
The following reply was made to PR pkg/34859; it has been noted by GNATS.

From: "Takayoshi Kochi" <takayoshi.kochi@gmail.com>
To: gnats-bugs@netbsd.org
Cc: taca@netbsd.org, gnats-admin@netbsd.org, pkgsrc-bugs@netbsd.org
Subject: Re: pkg/34859: ruby with native pthread broken
Date: Fri, 20 Oct 2006 18:39:58 +0900

 Hi Kambe-san,
 
 >  > >Environment:
 >  > NetBSD 3.1_RC4 on i386
 >  > >Description:
 >  > Trying to use bsfilter (http://bsfilter.org/) with ruby built from
 >  > pkgsrc doesn't work correctly.
 >  >
 >  > This PR might belong to both pkg and lib(pthread).
 >  Are you running this machine on VMWare as like PR pkg/34857?
 
 No, 3.1RC_4 is running on a real i386 (Celeron 800MHz) machine.
 
 I doubted my build was somewhat corrupt (like pkg/34857, though this
 is not -current),
 so today I refreshed my userland from daily snapshot build:
 
 ftp://ftp.netbsd.org/pub/NetBSD-daily/netbsd-3-1-RC4/200610171324Z/
 
 and tried the ruby with native pthread (rebuilt), but this happened again.
 So this is reproducible on my environment.
 
 To add some facts I learned, the ruby error "comparison of Fixnum with
 Float failed"
 appears pretty random.  I.e. sometimes this error occurs during
 "bsfilter --update"
 but sometimes not under the same condition.
 
 Today I tried this on -current on i386/amd64 (as I can build ruby on
 -current now ;)
 but this problem didn't happen (though the testing is limited).
 
 Now I am thinking of a possibility of a pthread locking problem that has been
 fixed after 3.0 but before -current.
 
 As the ruby with native pthread worked correctly on -current, I think
 this is not
 really a problem of ruby itself now.
 
 Kambe-san, do you have any idea whether "comparison of Fixnum with Float failed"
 can happen on `min' method for an array?
 
 -- 
 Takayoshi Kochi