«

Hi,

I attempted to run ruby gems on my 'new' Sun Ultra 5 machine which
runs OpenBSD 3.5. Here is the output:

# gem install rails
Attempting local installation of 'rails'
Local gem file not found: rails*.gem
Attempting remote installation of 'rails'
longjmp botch.
Illegal instruction (core dumped)

I have posted the core dump to
http://wmoxam.freeshell.org/files/ruby.core but I'm not sure if it's
of much use. If anyone can help me out with this problem it would be
greatly appriciated.

-- Wes

Wes Moxam wrote:
> Hi,
>
> I attempted to run ruby gems on my 'new' Sun Ultra 5 machine which
> runs OpenBSD 3.5. Here is the output:
>
> # gem install rails
> Attempting local installation of 'rails'
> Local gem file not found: rails*.gem
> Attempting remote installation of 'rails'
> longjmp botch.
> Illegal instruction (core dumped)
>
> I have posted the core dump to
> http://wmoxam.freeshell.org/files/ruby.core but I'm not sure if it's
> of much use. If anyone can help me out with this problem it would be
> greatly appriciated.
>
> -- Wes

this is not a problem with RubyGems, its a problem with ruby. There are
various minor code changes between BSDs. Before I attempt to pass this
core through my debugger(not junk gdb), which ruby version are you
using? Please make sure this is the latest before I spend time. Make you
you try it on the ruby-stable snapshot.

--dross

On Wed, 29 Sep 2004 10:48:03 +0900, David Ross <> wrote:
>
>
> Wes Moxam wrote:
> > Hi,
> >
> > I attempted to run ruby gems on my 'new' Sun Ultra 5 machine which
> > runs OpenBSD 3.5. Here is the output:
> >
> > # gem install rails
> > Attempting local installation of 'rails'
> > Local gem file not found: rails*.gem
> > Attempting remote installation of 'rails'
> > longjmp botch.
> > Illegal instruction (core dumped)
> >
> > I have posted the core dump to
> > http://wmoxam.freeshell.org/files/ruby.core but I'm not sure if it's
> > of much use. If anyone can help me out with this problem it would be
> > greatly appriciated.
> >
> > -- Wes
>
> this is not a problem with RubyGems, its a problem with ruby. There are
> various minor code changes between BSDs. Before I attempt to pass this
> core through my debugger(not junk gdb), which ruby version are you
> using? Please make sure this is the latest before I spend time. Make you
> you try it on the ruby-stable snapshot.
>
> --dross
>
>

Hi David,

I'm using the standard OpenBSD package which is version ruby-1.8.1. I
can download and compile a CVS version if you like.

-- Wes

Wes Moxam wrote:

>On Wed, 29 Sep 2004 10:48:03 +0900, David Ross <> wrote:
>
>
>>Wes Moxam wrote:
>>
>>
>>>Hi,
>>>
>>>I attempted to run ruby gems on my 'new' Sun Ultra 5 machine which
>>>runs OpenBSD 3.5. Here is the output:
>>>
>>># gem install rails
>>>Attempting local installation of 'rails'
>>>Local gem file not found: rails*.gem
>>>Attempting remote installation of 'rails'
>>>longjmp botch.
>>>Illegal instruction (core dumped)
>>>
>>>I have posted the core dump to
>>>http://wmoxam.freeshell.org/files/ruby.core but I'm not sure if it's
>>>of much use. If anyone can help me out with this problem it would be
>>>greatly appriciated.
>>>
>>>-- Wes
>>>
>>>
>>this is not a problem with RubyGems, its a problem with ruby. There are
>>various minor code changes between BSDs. Before I attempt to pass this
>>core through my debugger(not junk gdb), which ruby version are you
>>using? Please make sure this is the latest before I spend time. Make you
>>you try it on the ruby-stable snapshot.
>>
>>--dross
>>
>>
>>
>>
>
>Hi David,
>
>I'm using the standard OpenBSD package which is version ruby-1.8.1. I
>can download and compile a CVS version if you like.
>
>-- Wes
>
>
I would try downloading the Ruby stable snapshot. 1.8.1 might have a bug
that was fixed in the stable CVS. I can't watch the CVS logs and CIA(the
CVS watcher bot on my freenode channel) all the time. This issue might
have already been addressed.

--dross

>>>>> "W" == Wes Moxam <> writes:

W> longjmp botch.

_setjmp()/_longjmp() is broken on OpenBSD/Sparc, replace it with
setjmp()/longjmp()

See [ruby-talk:96816] and [ruby-talk:96957]

You have 2 occurence of _setjmp() to change in ruby


Guy Decoux

ts wrote:
>>>>>>"W" == Wes Moxam <> writes:
>
>
> W> longjmp botch.
>
> _setjmp()/_longjmp() is broken on OpenBSD/Sparc, replace it with
> setjmp()/longjmp()
>
> See [ruby-talk:96816] and [ruby-talk:96957]
>
> You have 2 occurence of _setjmp() to change in ruby
>
>
> Guy Decoux

Obviously its broken. The libraries are slightly different. Do you know
if a patch been commited?

--dross

>>>>> "D" == David Ross <> writes:

D> Obviously its broken. The libraries are slightly different. Do you know
D> if a patch been commited?

I don't know who is the maintainer for openbsd/ruby


Guy Decoux

On Wed, 29 Sep 2004 22:17:25 +0900, ts <> wrote:
> >>>>> "D" == David Ross <> writes:
>
> D> Obviously its broken. The libraries are slightly different. Do you know
> D> if a patch been commited?
>
> I don't know who is the maintainer for openbsd/ruby
>
> Guy Decoux
>

I tried ruby-1.8.2-preview2 and had the same problem. I also tried
compiling the latest code from the cvs server, but had problems
compiling it (yacc was complaining about parse.y).

I will try replacing _setjmp()/_longjmp() as suggested and
re-compile. I'll also try sending a message to the package maintainer
for OpenBSD/Sparc so that this can be corrected.

-- Wes

On Wed, Sep 29, 2004 at 07:35:19PM +0900, ts wrote:
> >>>>> "W" == Wes Moxam <> writes:
>
> W> longjmp botch.
>
> _setjmp()/_longjmp() is broken on OpenBSD/Sparc, replace it with
> setjmp()/longjmp()
>
> See [ruby-talk:96816] and [ruby-talk:96957]
>
> You have 2 occurence of _setjmp() to change in ruby
>

This is no longer true. The sparc/sparc64 versions of _setjump/_longjump
are fixed in OpenBSD 3.6. The change did not make it back to 3.5-stable so
you need to patch or update to 3.6/-current.

--
:wq Claudio

Claudio Jeker wrote:

>On Wed, Sep 29, 2004 at 07:35:19PM +0900, ts wrote:
>
>
>>>>>>>"W" == Wes Moxam <> writes:
>>>>>>>
>>>>>>>
>>W> longjmp botch.
>>
>> _setjmp()/_longjmp() is broken on OpenBSD/Sparc, replace it with
>> setjmp()/longjmp()
>>
>> See [ruby-talk:96816] and [ruby-talk:96957]
>>
>> You have 2 occurence of _setjmp() to change in ruby
>>
>>
>>
>
>This is no longer true. The sparc/sparc64 versions of _setjump/_longjump
>are fixed in OpenBSD 3.6. The change did not make it back to 3.5-stable so
>you need to patch or update to 3.6/-current.
>
>
>
Oh, good good. So there needs to be a patch for OpenBSD 3.5 and
whichever are incompatible.

--dross