Re: custom model rows-reordered marshal

[Kevin Ryde <user42 zip com au>]

Torsten Schoenfeld <kaffeetisch gmx de> writes:
>
> So I applied your patch.

I suspect it's still not quite right, I've got doubts about the ST() in
the ->get_value() supplied-columns case, the same as was wrong with what
I first tried for g_object_get.

> I think the iter_n_children case was fixed by moving the code that
> might end up overwriting the stack above the code that modifies the
> local stack pointer.

I can't remember at all now what I was looking at there.

> I adjusted the test case slightly and now see the stack movement too.

A tricky bit it trying to write something that doesn't use up the stack
too soon.  If it's already grown then obviously you don't get a problem.
Might have to stick the subtler test cases in separate files to be
confident of provoking the problem.