Command in backticks hangs? (OS X, 1.8.6-p36)

[Tim Hunter]

I'm running ruby 1.8.6 (2007-06-07 patchlevel 36) [powerpc-darwin8.9.0]

I have a script (the OS X installer script for RMagick) that I've been
using for some months now. Today I discovered that, in this script only,
a command in backticks hangs:

im_version = `Magick-config --version` rescue nil

The command works normally when I enter it on the command line. The
statement works normally when I try it in irb. This statement works
normally as a simple one line Ruby script.

It is possible that this is the first time I've run the script since I
upgraded to this version of Ruby.

Has anybody else encountered this problem?

 

[Tim Hunter]

I'm running ruby 1.8.6 (2007-06-07 patchlevel 36) [powerpc-darwin8.9.0]

I have a script (the OS X installer script for RMagick) that I've been
using for some months now. Today I discovered that, in this script
only, a command in backticks hangs:

im_version = `Magick-config --version` rescue nil

The command works normally when I enter it on the command line. The
statement works normally when I try it in irb. This statement works
normally as a simple one line Ruby script.

It is possible that this is the first time I've run the script since I
upgraded to this version of Ruby.

Has anybody else encountered this problem?

Following up to myself: I uninstalled 1.8.6 patchlevel 36 and replaced
it with 1.8.6 patchlevel 0 downloaded from RubyForge. The script works
normally now. I suppose I should get energetic and work on a small
reproduction of the problem.

 

[James Edward Gray II]

Has anybody else encountered this problem?

Yes. This is the bug that hosed the Ruby Quiz software:

http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/256794

That's the second time I have run into the bug too, unfortunately.
It's a pretty big issue with that release.

It is fixed in the HEAD of Ruby's 1.8 branch, if that helps.

James Edward Gray II

 

[Tim Hunter]

Yes. This is the bug that hosed the Ruby Quiz software:

http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-talk/256794

That's the second time I have run into the bug too, unfortunately.
It's a pretty big issue with that release.

It is fixed in the HEAD of Ruby's 1.8 branch, if that helps.

James Edward Gray II

Thanks! I just hope it's not broken in the version that will ship with
Leopard.

 

[James Edward Gray II]

Thanks! I just hope it's not broken in the version that will ship
with Leopard.

Yeah, I fear that myself. It would hurt the use of Ruby as a
scripting language in that release a lot.

James Edward Gray II

 

[James Edward Gray II]

The script works normally now. I suppose I should get energetic and
work on a small reproduction of the problem.

If anyone has a minimal script that shows this error, please share
it. I've been in contact with Apple and Leopard is suppose to ship
with Ruby 1.8.6-p36 which is where I found the issue. They are
willing to patch it though, if we can show the problem.
Unfortunately, both scripts I ran into it with are very non-trivial.

James Edward Gray II

 

[Tim Hunter]

If anyone has a minimal script that shows this error, please share
it. I've been in contact with Apple and Leopard is suppose to ship
with Ruby 1.8.6-p36 which is where I found the issue. They are
willing to patch it though, if we can show the problem.
Unfortunately, both scripts I ran into it with are very non-trivial.

James Edward Gray II

Oh, great. I re-installed 1.8.6-p36 and the script runs normally now.

Sigh. Sometimes I hate software.

 

[James Edward Gray II]

Oh, great. I re-installed 1.8.6-p36 and the script runs normally now.

Sigh. Sometimes I hate software.

I'm pretty sure the issue is related to threading. It would always
show up for me after I had been fetching Web pages or sending some
email using the networking libraries.

James Edward Gray II

 

[Bill Kelly]

I'm pretty sure the issue is related to threading. It would always
show up for me after I had been fetching Web pages or sending some
email using the networking libraries.

Something like this was brought up on ruby-core last week:
http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-core/11505

``1.8.6 thread support was broken in bad ways. It stayed for three months

1.8.6-p36 is broken w.r.t. compilation of extensions and has a thread bug.
The bug is much minor than in 1.8.6, but it is nonetheless a bug in a
very "core" part of the standard library.''

It seems to me if apple is willing to apply a patch, we ought to
maybe bring this discussion over to ruby-core ASAP so that whoever
knows the details can co-ordinate with apple... It sounds like
there is the potential to avert a disaster here...


Regards,

Bill

 

[James Edward Gray II]

Something like this was brought up on ruby-core last week:
http://blade.nagaokaut.ac.jp/cgi-bin/scat.rb/ruby/ruby-core/11505

``1.8.6 thread support was broken in bad ways. It stayed for three
months

1.8.6-p36 is broken w.r.t. compilation of extensions and has a
thread bug. The bug is much minor than in 1.8.6, but it is
nonetheless a bug in a very "core" part of the standard library.''

Yes, both myself and an Apple employee were trying to follow that
message train to find example error code and see where the issue was
resolved in Subversion. We may have located the relevant patch by
Nobu in the bug tracker, but that's about it.

James Edward Gray II