Re: [Denemo-devel] Darwin crash

[Richard Shann]

On Wed, 2015-10-21 at 18:26 -0500, Jeremiah Benham wrote:
> 
> On Wed, Oct 21, 2015 at 3:48 PM, Richard Shann
> <address@hidden> wrote:
[...]
>         
>         You could send me your Default.shortcuts and/or the value of
>         cursor_num
>         on entry to the function assign_cursor().
>         
> 
> 
> I uncommented this line in mousing.c:
> g_print("Storing cursor %d for state 0x%x in hash table %p\n",
> cursor_num, state, Denemo.map->cursors );
> 
> and it printed this:
> Storing cursor 116 for state 0x104 in hash table 0x20ae1a0
> Storing cursor -1 for state 0x102 in hash table 0x20ae1a0
> 
> 
> I am attaching my Default.shortcuts.

Well, I was right, the second entry in your Default.shortcuts is setting
a cursor for 
       <cursor-binding>
          <state>102</state>
          <cursor>-1</cursor>
        </cursor-binding>

and cursor number -1 is presumably out of range. Looking at the code I
see it has been causing trouble in the past and is suffering from
bit-rot. I've increasingly introduced hard-wired mouse operations over
the past few years (double-click for information, right-click for edit,
drag for selection extension ...) and trying out the interface to the
cursor shape just now in the Command Center it seems to be quite broken.

I think it would be good to drop this stuff for release 2.0.0 as I don't
think it is being used.

Will you be able to merge into the stable-2.0.0 branch when I have
finished doing that? I realize it means yet more delay in getting this
release out, but I would like it to be a much more finished release than
earlier ones have been.

Richard