[Bf-committers] [Bug 17457] Corrupted input file?

[Brecht Van Lommel]

Hi,

As far as I can see mr->verts should not be NULL here, it looks like the
multires data got corrupted somehow. Which operation made this happen is
the question then, adding NULL pointer checks is not really a solution
as it will likely crash in other places anyway.

The multires summer code project is intended to replace this code in the
future. It would be good to have multires bugs fixed for 2.48 of course,
but I would not spend too much time on this one.

Brecht.

On Fri, 2008-10-03 at 02:52 -0400, Jingyuan Huang wrote:
> Hi everyone,
> I'm new to development for Blender. I looked at the bug tracker and
> would like to start with some bug fixings. I'm looking at bug #17457
> and found that after the input file is loaded, a mesh object would
> have multires info, but no vertices. Adding additional output to
> link_global in source/blender/blenloader/intern/readfile.c gives me
> the following output, where FORMAT is FileGlobal->subvstr (I don't
> know what it represents), multires is m->mr, and verts is m->mr->verts
> if m->mr is not NULL.

> A simple workaround fix would be the following, though I'm really not
> satisfied with this kind of fixing. What could be the possible causes
> for multires having no verts? I'd like to get a couple of pointers
> before I poke the code a bit more. Thanks a lot.
>