Roll skia/third_party/skcms d7c7dd8b955e..ce4270691346 (1 commits)

https://skia.googlesource.com/skcms.git/+log/d7c7dd8b955e..ce4270691346

git log d7c7dd8b955e..ce4270691346 --date=short --no-merges --format='%ad %ae %s'
2019-08-23 mtklein@google.com Two steps as good as three, maybe better?

Created with:
  gclient setdep -r skia/third_party/skcms@ce4270691346

The AutoRoll server is located here: https://autoroll.skia.org/r/skcms-skia-autoroll

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.


CQ_INCLUDE_TRYBOTS=luci.chromium.try:linux-blink-rel

Bug: None
Change-Id: I5cd27cce50a3529243b1d46fb0fb8f6c3ddab235
TBR=egdaniel@google.com,mtklein@google.com
TBR=egdaniel@google.com,mtklein@google.com
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/236874
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>

third_party/skcms/skcms.cc

@@ -1672,7 +1672,8 @@ static bool fit_nonlinear(const skcms_Curve* curve, int L, int N, skcms_Transfer
     // No matter where we start, dx should always represent N even steps from 0 to 1.
     const float dx = 1.0f / (N-1);
 
-    for (int j = 0; j < 3/*TODO: tune*/; j++) {
+    // As far as we can tell, 1 Gauss-Newton step won't converge, and 3 steps is no better than 2.
+    for (int j = 0; j < 2; j++) {
         // These extra constraints a >= 0 and ad+b >= 0 are not modeled in the optimization.
         // We don't really know how to fix up a if it goes negative.
         if (P[1] < 0) {

third_party/skcms/version.sha1

@@ -1 +1 @@
-d7c7dd8b955e9633da5ba763bb707854709d3033
+ce4270691346d0df87116ab8a118877102fe3137