Re: [Rubycocoa-talk] How to debug RubyCocoa?

[Pierce T. W. I. <pi...@tw...>]

On Nov 29, 2007, at 12:24 PM, Chris McGrath wrote:

> Puts $RUBYCOCOA_DEBUG =3D true in your app somewhere and have a look =
at
> the output, that should give you more of a pointer to what's actually
> going wrong.

Yeah, I've done that already, and posted the results to this list =20
before.

It's buserroring in install_ovmix_class_methods after building an =20
OVMIX class.

Here's the console output:

2007-11-30 09:23:01.409 Frictionless[8610:813] OVMIX : Registered Ruby =20=

method by selector 'tableView:willDisplayCell:forTableColumn:row:' =20
types 'v24@0:4@8@12@16i20'
2007-11-30 09:23:01.413 Frictionless[8610:813] OVMIX : retrieving =20
closure imp for method type 'v24@0:4@8@12@16@20'
2007-11-30 09:23:01.414 Frictionless[8610:813] DATACNV : decoding =20
method encoding 'v24@0:4@8@12@16@20' manually
2007-11-30 09:23:01.414 Frictionless[8610:813] DATACNV : retval -> v
2007-11-30 09:23:01.415 Frictionless[8610:813] DATACNV : argc -> 6
2007-11-30 09:23:01.416 Frictionless[8610:813] DATACNV : arg[0] -> @
2007-11-30 09:23:01.416 Frictionless[8610:813] DATACNV : arg[1] -> :
2007-11-30 09:23:01.417 Frictionless[8610:813] DATACNV : arg[2] -> @
2007-11-30 09:23:01.418 Frictionless[8610:813] DATACNV : arg[3] -> @
2007-11-30 09:23:01.418 Frictionless[8610:813] DATACNV : arg[4] -> @
2007-11-30 09:23:01.419 Frictionless[8610:813] DATACNV : arg[5] -> @
2007-11-30 09:23:01.420 Frictionless[8610:813] LIBFFI : make closure =20
argc 6
2007-11-30 09:23:01.420 Frictionless[8610:813] LIBFFI : arg[0] -> =20
ffi_type 0xa011e034
2007-11-30 09:23:01.421 Frictionless[8610:813] LIBFFI : arg[1] -> =20
ffi_type 0xa011e034
2007-11-30 09:23:01.422 Frictionless[8610:813] LIBFFI : arg[2] -> =20
ffi_type 0xa011e034
2007-11-30 09:23:01.423 Frictionless[8610:813] LIBFFI : arg[3] -> =20
ffi_type 0xa011e034
2007-11-30 09:23:01.423 Frictionless[8610:813] LIBFFI : arg[4] -> =20
ffi_type 0xa011e034
2007-11-30 09:23:01.424 Frictionless[8610:813] LIBFFI : arg[5] -> =20
ffi_type 0xa011e034
2007-11-30 09:23:01.424 Frictionless[8610:813] OVMIX : Registered Ruby =20=

method by selector 'outlineView:willDisplayCell:forTableColumn:item:' =20=

types 'v24@0:4@8@12@16@20'
Program received signal:  =93EXC_BAD_ACCESS=94.

The above method is the last method in ActionBrowser.rb, a subclass of =20=

NSObject. So its registered a bunch of methods and I'm guessing its =20
building a class based on the call to RBObjcDerivedClassNew in the =20
backtrace.

Here's the stack backtrace:

#0	0x0005f713 in install_ovmix_class_methods
#1	0x000664dc in RBObjcDerivedClassNew
#2	0x00065f22 in ocobj_s_new_with_class_name
#3	0x000d10b7 in rb_with_disable_interrupt
#4	0x000da636 in rb_eval_string_wrap
#5	0x000db24a in rb_eval_string_wrap
#6	0x000d84c5 in rb_eval_string_wrap
#7	0x000d8e92 in rb_eval_string_wrap
#8	0x000d6781 in rb_eval_string_wrap
#9	0x000dae3c in rb_eval_string_wrap
#10	0x000db24a in rb_eval_string_wrap
#11	0x000d84c5 in rb_eval_string_wrap
#12	0x000dae3c in rb_eval_string_wrap
#13	0x000db24a in rb_eval_string_wrap
#14	0x000dbccd in rb_respond_to
#15	0x000dbdb6 in rb_funcall
#16	0x000c81ab in rb_class_inherited
#17	0x000d9e0e in rb_eval_string_wrap
#18	0x000e6a47 in rb_load
#19	0x000e70a8 in rb_require_safe
#20	0x000da636 in rb_eval_string_wrap
#21	0x000db24a in rb_eval_string_wrap
#22	0x000d84c5 in rb_eval_string_wrap
#23	0x000e6a47 in rb_load
#24	0x000e70a8 in rb_require_safe
#25	0x000da636 in rb_eval_string_wrap
#26	0x000db24a in rb_eval_string_wrap
#27	0x000d84c5 in rb_eval_string_wrap
#28	0x000de658 in rb_thread_trap_eval
#29	0x000df506 in rb_yield
#30	0x000c010d in rb_ary_each
#31	0x000da636 in rb_eval_string_wrap
#32	0x000db24a in rb_eval_string_wrap
#33	0x000d84c5 in rb_eval_string_wrap
#34	0x000d7156 in rb_eval_string_wrap
#35	0x000dae3c in rb_eval_string_wrap
#36	0x000db24a in rb_eval_string_wrap
#37	0x000d84c5 in rb_eval_string_wrap
#38	0x000e7388 in rb_load_protect
#39	0x000e73b9 in ruby_exec
#40	0x000e73e5 in ruby_run
#41	0x0006488c in RBApplicationMain
#42	0x000029fd in main at main.m:13

But with unfamiliar code, I'd rather have source access to the =20
install_ovmix_class_methods so that I can report a solution to the =20
rubycocoa developers, as it what pointer in particular is bad...

  Pierce

P.S.

  Source code to app available at =
http://www.twinforces.com/frictionless/=20
  if anyone wants to look at it. All you have to do is start it up on =20=

Leopard to reproduce the crash.