Hi Mario has been working on the MingW build and has been encountering a lot of crashes. It turns out it''s not specific to MingW but rather to the new versions of Ruby, 1.8.7. With 1.8.7-preview2 on Linux too the whole thing becomes very unstable. It seems that the ruby core developers have made some major changes to the way that GC is handled, and this has broken the GC marking stuff in wxRuby. I''m not sure at the moment whether the problem is in our code, or perhaps a problem with SWIG. Anyway, my plan is to proceed with releasing wxRuby 1.9.6 in the next few weeks, but warn that it''s not yet usable with 1.8.7. alex
On Sun, Apr 20, 2008 at 12:42 PM, Alex Fenton <alex at pressure.to> wrote:> Hi > > Mario has been working on the MingW build and has been encountering a > lot of crashes. It turns out it''s not specific to MingW but rather to > the new versions of Ruby, 1.8.7. With 1.8.7-preview2 on Linux too the > whole thing becomes very unstable. > > It seems that the ruby core developers have made some major changes to > the way that GC is handled, and this has broken the GC marking stuff in > wxRuby. I''m not sure at the moment whether the problem is in our code, > or perhaps a problem with SWIG. > > Anyway, my plan is to proceed with releasing wxRuby 1.9.6 in the next > few weeks, but warn that it''s not yet usable with 1.8.7. > > alexIt''s also worth noting, that wxRuby isn''t the only project having problems with 1.8.6 to 1.8.7 changes. A lot of other projects are having similar problems. Notably Rubinius'' rubyspecs furballed under the new 1.8.7 release. So keep this in mind, hold off on the 1.8.7 release for a while. -- Mario Steele http://www.trilake.net http://www.ruby-im.net http://rubyforge.org/projects/wxruby/ http://rubyforge.org/projects/wxride/ http://rubyforge.org/projects/vwmc/ -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/wxruby-development/attachments/20080420/47a22620/attachment.html
Mario Steele wrote:> It''s also worth noting, that wxRuby isn''t the only project having > problems with 1.8.6 to 1.8.7 changes. A lot of other projects are > having similar problems. Notably Rubinius'' rubyspecs furballed under > the new 1.8.7 release. So keep this in mind, hold off on the 1.8.7 > release for a while.That''s reassuring if depressing. It''s taken myself and others a huge amount of time to get the GC / memory management sorted to the level of stability we have today in wxRuby + ruby 1.8.6. The compile... wait... wait... run.. click... crash... gdb...wait.. wait... tweak... compile... cycle that it takes to get there is about the least fun thing I know. Given that 1.8.7 is a dead-end release, with 2.0 in the works, I don''t intend to spend any more time on it. Mario, if you or someone else can figure it out in a way that''s backward compatible at least to 1.8.6 not p114 (as this is the standard version on OS X), great. alex
As much as I hate to say it, and as depressing as this is going to be.... it may very well be that 2.0 will have the same thing in it, as far as the rb_bug() which originally started this quest to fix it. With as much as they back ported from 2.0, to 1.8, it may be something that is 2.0 related, and we''ll end up being in the same boat as we are now. Do realize Alex, I''m not trying to get you down. Just a sad, but very possible truth of the matter. On Sun, Apr 20, 2008 at 2:23 PM, Alex Fenton <alex at pressure.to> wrote:> Mario Steele wrote: > > It''s also worth noting, that wxRuby isn''t the only project having > > problems with 1.8.6 to 1.8.7 changes. A lot of other projects are > > having similar problems. Notably Rubinius'' rubyspecs furballed under > > the new 1.8.7 release. So keep this in mind, hold off on the 1.8.7 > > release for a while. > That''s reassuring if depressing. > > It''s taken myself and others a huge amount of time to get the GC / > memory management sorted to the level of stability we have today in > wxRuby + ruby 1.8.6. The compile... wait... wait... run.. click... > crash... gdb...wait.. wait... tweak... compile... cycle that it takes to > get there is about the least fun thing I know. > > Given that 1.8.7 is a dead-end release, with 2.0 in the works, I don''t > intend to spend any more time on it. Mario, if you or someone else can > figure it out in a way that''s backward compatible at least to 1.8.6 not > p114 (as this is the standard version on OS X), great. > > alex > > > > _______________________________________________ > wxruby-development mailing list > wxruby-development at rubyforge.org > http://rubyforge.org/mailman/listinfo/wxruby-development >-- Mario Steele http://www.trilake.net http://www.ruby-im.net http://rubyforge.org/projects/wxruby/ http://rubyforge.org/projects/wxride/ http://rubyforge.org/projects/vwmc/ -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/wxruby-development/attachments/20080420/7fd61f32/attachment-0001.html
Mario Steele wrote:> As much as I hate to say it, and as depressing as this is going to > be.... it may very well be that 2.0 will have the same thing in it, as > far as the rb_bug() which originally started this quest to fix it. > With as much as they back ported from 2.0, to 1.8, it may be something > that is 2.0 related, and we''ll end up being in the same boat as we are > now. > > Do realize Alex, I''m not trying to get you down. Just a sad, but very > possible truth of the matter.Hehe, I''m less gloomy and more grumpy. You''re probably right, although at first glance it worked OK with 1.9.0. But at least with a version 2.0 you expect some pain, and get a bit more than Enumerable#minmax_by in return. Given the big internal changes, 1.8.7 won''t necessarily be a good guide to problems we might hit with 2.0. With some luck someone will take an interest in improving SWIG/Ruby and much of it will be taken from there. cheers alex