On Sun, Jan 21, 2001 at 07:07:00PM +0100, gerard patel wrote:
> At 03:53 PM 1/21/01 +0000, you wrote:
>
> >Yes I wondered about that. The problem is, what is the Right Thing - fail the
> >BitBlt operation? Also Alexandre mentioned a few days ago that he was not keen
> >on rippling errors back up the function chain if nothing useful could be done at
> >the top-level.
>
> In this case it's an application bug so it's certainly a better option than
> exiting Wine.
>
> >On futher digging I wonder if I've found the root of the problem. The Win32 API
> >says that DeleteObject() does not delete an object which is currently
> >selected into a DC, and returns 0 in that case. Wine doesn't seem to
> >carry out that check, and from the trace I posted earlier the DeleteObject() on
> >the in-use bitmap returned 1.
>
> I missed that :-(
>
> You can try the following patch :
> [Corrected] Poor Man's GDI Object Refcount
>
> posted on 2000/01/15
>
> It is addressing these issues.
>
> But it will probably never be committed to cvs as it is, since it's (at least)
>failing
> to special case regions :-/, so if you are using it your app may soon run out of
> gdi space.
Finally somebody mentions that this has already been discussed... ;-)
I just added bug #165 for this, BTW.
Andreas Mohr
