[Java-gnome-developer] Re: Hold the patch

[Jean v. W. <je...@sm...>]

Hi Julian,

On Thu, 22 Feb 2001, Julian Fitzell wrote:

> Hmm... I don't seem to be getting subscribed to the list so I'm not sure if
> that email I sent went through or not... but I am having some seg faulting

Yes it did come through.  I was just busy with something so could not reply.

> problems with the LibGlade stuff (at least on Solaris - it's related to
> pointers and I think it works ok on Linux) so once I get these resolved I
> will submit the patch again with both sets of changes.
> 
> Julian Fitzell

I incorporated Avi's initial patch into the tree.  It worked fine except on
three points:

1) Certain cases created a core dump/crash so I had to remove a few lines
of code.

2) There was no GNOME support.

3) We moved the packages to gnu.gtk, gnu,gdk etc.

2 and 3 was trival to add.  An explenation on 1 follows:

The way that the callback mechanism works is: it expects that the object
registering the callback will also handle the callback (surprise,
surprise).  The way Avi's code handled it was correct except for this
miscalculation.  When using LibGlade, the physical object does not handle
the callback but the Object that calls the gnu.glade.LibGlade class does.
This meant that for certain callbacks the code thought that, that object
handled the callback rather than the Object that reads the file, resulting
in a crash.  Unfortunately the crash is caused in such a way that I have
not yet found a solution for it.

Some more information:

Unfortunately the LibGlade code is not very resilient currently and will
crash when it wants to bind a callback to a method that does not exist.

My suggestions:

Use the CVS code rather than  0.5 (if you are not already doing it), this
will hopefully solve the problem of the one type of crash.

Secondly make sure that all the methods you want to bind too exists.

Three, look at the examples in src/examples/libglade and
src/examples/libglade-gnome.

Please feel free to contact me for more information.

Regards
Jean

(PS. For interest sake:  diff -u diffs are easier to read and understand
than the ordinary diffs. --My personal oppinion)