Re: Glib::Object::Subclass bug? or Misunderstanding?

[muppet <scott asofyet org>]

On Mar 20, 2006, at 10:18 PM, James Muir wrote:

> muppet wrote:
> > What upgrade was it? Was glib-2.0 upgraded without rebuilding  
> > Glib? There was a rash of "bugs" recently related to botched  
> > upgrades...
> >
> > I can't reproduce this. I've tried
> >
> > - Gnome2::Canvas 1.002.
> > - libgnomecanvas 2.6.0
> > - libgnomecanvas 2.11.1
> >
> > - Glib 1.105
> > - Glib 1.120
> > - glib 2.4.8
> > - glib 2.10.2
> Another error in communication on my part. I performed a clean  
> install of Zenwalk 2.2.
>
> I tried inserting the default GET_PROPERTY into the modules that  
> use Glib::Object::Subclass and I got further. Some objects were  
> displayed on the canvas, but I'm now getting:
>
> *** unhandled exception in callback:
> *** Can't locate object method "get_default_value" via package  
> "Glib::Param::Boxed" at ../lib/Gtk2/Ex/MindMapView/Item.pm line 227.
> *** ignoring at example4.pl line 68.

That's because you can't specify a default value for a Boxed  
parameter type.  The code in the XS implementation does the  
equivalent of "$pspec->can ('get_default_value') && return $pspec- 
>get_default_value".  The documentation omits that for brevity and  
clarity.  *cough*

If you're planning to use integer values for that property, use  
Glib::Param::Int instead.

> and also I ran the Glib and Gtk2 regression tests. It looks to me  
> like the "f.t" Glib test is confirming that I have problem with  
> Glib::Object::Subclass. Maybe you can make more of it. Here's the  
> Glib test results:

[...]

> t/f........................
> # Failed test 'and they have values'
> # in t/f.t at line 111.
> # Structures begin differing at:
> # $got->[0] = undef
> # $expected->[0] = 'Joe'
>
> # Failed test 'and they have values'
> # in t/f.t at line 168.
> # Structures begin differing at:
> # $got->[0] = undef
> # $expected->[0] = 'Joe'
> # Looks like you failed 2 tests of 42.
> dubious
> Test returned status 2 (wstat 512, 0x200)
> DIED. FAILED tests 13, 42
> Failed 2/42 tests, 95.24% okay

Please try with Glib 1.105.  I would swear till i was blue in the  
face that this is a problem we fixed in the last couple of months,  
but i can't remember the details.


> t/GtkTreeView................dubious
> Test returned status 0 (wstat 11, 0xb)
> DIED. FAILED tests 112-137
> Failed 26/137 tests, 81.02% okay

That's somewhat worrying...


> Any suggestions on how to proceed with debugging this problem would  
> be appreciated.

Try figuring out why the code is skipped.  I can't give many more  
details because i can't reproduce it.

--
muppet <scott at asofyet dot org>