'static const' means, there must be at most one of these, and initialize it at
compile time if possible or runtime if necessary. This leads to unexpected
code execution, and TSAN* will complain about races on the guard variables.
Generally 'constexpr' or 'const' are better choices. Neither can cause races:
they're either intialized at compile time (constexpr) or intialized each time
independently (const).
This CL prefers constexpr where possible, and uses const where not. It even
prefers constexpr over const where they don't make a difference... I want to have
lots of examples of constexpr for people to see and mimic.
The scoped-to-class static has nothing to do with any of this, and is not changed.
* Not yet on the bots, which use an older TSAN.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2300623005
Review-Url: https://codereview.chromium.org/2300623005
- Switch to portable typeface, to avoid per-platform diffs in the labels
- Stretch out the image slightly, to avoid overlap with larger font
- Fix several tests that were no longer matching ground-truth. We treat
SkColor as sRGB, so 0x7F is incorrect, for example. Several of the
XferMode tests had similar math errors now. Computed new values, and
verified that they work as expected.
- Removed a couple tests that were mixing color and alpha in funny ways
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2288053003
Review-Url: https://codereview.chromium.org/2288053003
Verify the rules that we're converging on for surfaces:
- For 8888, we only support sRGB-like gamma, or no color space at all.
- For F16, we require a color space, with linear gamma.
- For all other formats, we do not support color spaces.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2270823002
Review-Url: https://codereview.chromium.org/2270823002
This seems to work well for miter and bevel joins with the resulting stroke and fill path remaining convex. There seems to be an issue with round joins where the outer generated shell is usually not convex. Without this CL the resulting stroke & filled paths are always concave.
Perf-wise (on Windows):
convex-lineonly-paths-stroke-and-fill bench
(in ms) w/o w/CL %decrease
8888 2.88 2.01 30.2
gpu 4.4 1.38 68.6
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2275243003
Review-Url: https://codereview.chromium.org/2275243003
Keep isOpaque as a convenience method -- many places really only need to
know that for optimization purposes (SrcOver -> Src, etc...).
In all the places where we pull data back out or convert to another
object and need to supply an SkImageInfo, we can avoid losing information
about premulness.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2250663002
Review-Url: https://codereview.chromium.org/2250663002
Reason for revert:
Broken Windows bot.
Original issue's description:
> Fix WIC encoder to support kJPEG_Type
>
> (1) Add support for kJPEG to WIC
> (2) Add encoding test.
> (3) Turn on WIC jpeg encoder on Windows and CG jpeg
> encoder on Mac.
>
> A follow-up may make Skia's encoders the default on all
> platforms. But, in order to do that, I think we need
> to write better encoding unit tests for CG and WIC.
>
> BUG=skia:3969
> BUG=skia:5632
> GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2245453002
>
> Committed: https://skia.googlesource.com/skia/+/b3a7ef1fc0adc24859d2498aee54d3ec2cbcac3aTBR=mtklein@google.com,mtklein@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=skia:3969
Review-Url: https://codereview.chromium.org/2246203002
(1) Add support for kJPEG to WIC
(2) Add encoding test.
(3) Turn on WIC jpeg encoder on Windows and CG jpeg
encoder on Mac.
A follow-up may make Skia's encoders the default on all
platforms. But, in order to do that, I think we need
to write better encoding unit tests for CG and WIC.
BUG=skia:3969
BUG=skia:5632
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2245453002
Review-Url: https://codereview.chromium.org/2245453002
In Firefox, we use SkCanvas::saveLayer in combination with a backdrop that initializes the layer to the background. When this is blended back onto background using transparency, where the source and destination pixel colors are the same, the resulting color after the blend is not preserved due to the lost precision mentioned above. In cases where this operation is repeatedly performed, this causes substantially noticeable differences in color as evidenced in this downstream Firefox bug report: https://bugzilla.mozilla.org/show_bug.cgi?id=1200684
In the test-case in the downstream report, essentially it does blend(src=0xFF2E3338, dst=0xFF2E3338, scale=217), which gives the result 0xFF2E3237, while we would expect to get back 0xFF2E3338.
This problem goes away if the blend is instead reformulated to effectively do (src*src_scale + dst*dst_scale)>>8, which keeps the intermediate precision during the addition before shifting it off.
This modifies the blending operations thusly. The performance should remain mostly unchanged, or possibly improve slightly, so there should be no real downside to doing this, with the benefit of making the results more accurate. Without this, it is currently unsafe for Firefox to blend a layer back onto itself that was initialized with a copy of its background.
BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=2097883002
CQ_INCLUDE_TRYBOTS=master.client.skia:Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Release-SKNX_NO_SIMD-Trybot
[mtklein adds...]
No public API changes.
TBR=reed@google.com
Review-Url: https://codereview.chromium.org/2097883002
