[PATCH 05/14] miPolyPoint bugfix: Check memory allocation before changing the GC.

[Jamey Sharp]

On Mon, May 10, 2010 at 8:18 AM, Adam Jackson <ajax at nwnk.net> wrote:
> On Sat, 2010-05-08 at 16:39 -0700, Jamey Sharp wrote:
>> miPolyPoint ought to leave the GC unchanged even if it fails.
>
> We have a new winner for the oldest-bug competition!  It's actually been
> like that since X11R1:
>
> -rw-r--r--. 1 ajax ajax 2817 1987-09-12 01:20 ddx/mi/mipolypnt.c

Sweet, I was vying for that spot. :-)

Where do you get history going back that far, and can we get it in
git-repo form? (I wonder if that would set a record for "oldest commit
in a git repository"?)

Regarding your comments on ChangeGC->dixChangeGC, I agree that the
CreateGC usage of NullClient is questionable. I didn't think about it
when I wrote that commit, but on later review I added a commit to fix
it, and CC'd Eamon. I think it's necessary to do access checks on the
XIDs in the value-list, not just the newly-created GC id.

Since Keith kept making me edit my patches ;-) maybe its best to
review from my personal repo now?

http://cgit.freedesktop.org/~jamey/xserver/log/?h=cleanups

In particular, the version of "Replace dixChangeGC with calls directly
to the right variant" that you reviewed looks a bit different after
his suggestions.

Jamey