On 2011/08/01 13:09:20, reed1 wrote: > Is the net result of this CL just that ...
13 years, 4 months ago
(2011-08-01 13:18:37 UTC)
#4
On 2011/08/01 13:09:20, reed1 wrote:
> Is the net result of this CL just that we added { } around the if?
No, previously the type mask was set after this:
if (inv == &tmp) {
*(SkMatrix*)this = tmp;
}
Meaning that when inv is initially this the typemask was never set on this.
Also, as Cary pointed out to me there are early exits in which case we would
have changed some values of inv without setting its typemask.
This fixes a bug in SampleDemo on the GPU on Windows (in Release only). The type
mask after an invert is coming out zero which causes a later matrix multiply to
think the mul is by identity when it is not. A shader is then applied with the
wrong matrix.
>
> LGTM
Issue 4819050: Fix setting of typemask in SkMatrix::invert when inv==this
(Closed)
Created 13 years, 4 months ago by bsalomon
Modified 13 years, 4 months ago
Reviewers: caryclark1, reed1
Base URL: http://skia.googlecode.com/svn/trunk/
Comments: 0