Comments
-
Greg Hurrell
Haven't seen this one before, but then, I have never tested against Ruby 1.9.2.
So this is occurring at a fairly low level by the looks of it, in the garbage collector of the Ruby runtime itself.
There are three possibilities that spring to mind here:
- you may have discovered a bug in the Ruby interpreter, like the "NOTE" you pasted says; this seems unlikely, but Ruby 1.9.2 is very new so it is not outside the realm of possibility
- you may have discovered a bug in Command-T; to be honest, this seems even less likely, as it has proven stable for some time on a range of different versions of Ruby
- you might be running into trouble if the version of Ruby that is present on your system and which you used when you built Command-T is not identical the version of Ruby that was used when Vim was built
So my advice to you would be to make sure that you aren't using mismatched versions of Ruby. Failing that, try a different version of Ruby to see if the problem is exclusive to 1.9.2.
Here I'm using the stock standard Ruby that comes with my system, which happens to be 1.8.7, as I see no call for running a bleeding-edge Ruby interpreter in my editing environment, which is principally for getting work done and not playing with the latest shiny baubles.
-
anonymous
I've been experiencing this (or a similar crash) for a while now, using Vim 7.3.50 and Ruby 1.87. After a few moments of using CommandT, Vim crashes with:
~/.vim/ruby/command-t/controller.rb:262: [BUG] gc_sweep(): unknown data type 0x18(0xb61bfa9c) ruby 1.8.7 (2010-08-16 patchlevel 302) [i686-linux] Vim: Caught deadly signal ABRT Vim: Finished.
The mentioned line/file/date type isn't actually authoritative, as it changes from crash to crash.
That being said, I believe this to be related to internal changes in Vim: Downgrading to 7.2.442 fixes all problems. On a related note, another plugin also written in Ruby, "lusty juggler", does not work on 7.3 at all. It normally is used for switching buffers, but does not detect any buffers using Vim 7.3.
-
Greg Hurrell
At a guess, the lusty juggler failure is most likely due to a known bug in the Ruby support in 7.3 where it always reports the current buffer number as zero. A workaround is possible; falling back to buffer name instead when the bug is present. The corresponding ticket for this issue in Command-T is ticket #1617. The bug's been reported upstream but as far as I know hasn't been dealt with yet.
Not sure if the crashes might be due to related bugs in 7.3's Ruby support. I've been watching the vim_dev mailing list and there have been a few Ruby-related patches fixing various bugs, but I don't know if any of those might be relevant here.
-
anonymous
I've been getting this since the 1.01 release. The second time I run Command T, Vim crashes with this error.
Vim: Caught deadly signal SEGV [11:45:07] Vim: Finished.
-
Greg Hurrell
If you look at what was changed from 1.0 to 1.0.1, nothing was altered that would explain a crash.
If you changed nothing else (ie. neither Vim nor the Ruby used when building either Vim or Command-T) then it is almost certainly a miscompilation. Have you tried blowing it away and re-building it? Have you tried going over the troubleshooting tips in the documentation which talk about checking/ensuring that you're using the exact same version of Ruby when building Command-T as was used when Vim was built?
-
Greg Hurrell
Closed ticket #1767 as a duplicate of this one.
-
Greg Hurrell
Just to keep things together, this forum thread may be referring to the same underlying issue.
-
Greg Hurrell
Closed ticket #1781 as a duplicate of this one.
-
Greg Hurrell
Seeing as all reported crashes have been due to incorrect compilation or bugs in Vim's Ruby support, going to mark this one as closed too.
-
Greg Hurrell
Status changed:
- From: new
- To: closed
-
buck
Wincent: Is there any upstream vim/ubuntu bug for this?
If my issue is really caused by incorrect ruby integration, then I'd imagine other ruby plugins would be having issues as well.
-
buck
FYI, here are the upstream bugs:
Debian (fixed/closed) http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=614336 Ubuntu (unassigned) https://bugs.launchpad.net/ubuntu/+source/vim/+bug/738423?comments=all
-
Greg Hurrell
I'm not sure if other Ruby plugins have issues, or only ones like Command-T that involve native C extensions (not many of those around).
-
buck
Wincent: I may be mistaken, but I don't believe that the C extension is involved. I wish I could read the stack trace better. Are you able to glean any clues from it?
Here's the change that fixed it, according to vim changelog. http://code.google.com/p/vim/source/detail?r=ee50cd1a3032
Add a comment
Comments are now closed for this issue.