skia2/dm
Mike Klein f34d0c79dc re-hash images once more artificially
Some of my early trybots from the first big re-hash CL produced .pngs
without embedded ICC profiles, and some produced .pngs with embedded
profiles, with the same hash.  These were uploaded to Gold, of course.

Where the trybot-without-profile images uploaded first, we see Rec.2020
colors uncorrected as if sRGB, typically rather dim.  Where we're still
producing the same image we see these profile-less images and it's very
confusing, downright misleading.  Makes it impossible to triage.

So, add a salt mechanism to make it easy to re-hash everything,
invalidating all the old hashes so new profile-full images get uploaded.

No-Tree-Checks: true
Change-Id: I51f2bc4612b8a4f38d162124341d27a8e94c6085
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/201702
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
2019-03-16 02:39:12 +00:00
..
DM.cpp re-hash images once more artificially 2019-03-16 02:39:12 +00:00
DMGpuTestProcs.cpp Move GrBackend enum to enum class. 2018-10-12 14:01:58 +00:00
DMJsonWriter.cpp add color_depth field in DM/Gold 2019-03-07 21:32:35 +00:00
DMJsonWriter.h add color_depth field in DM/Gold 2019-03-07 21:32:35 +00:00
DMSrcSink.cpp rework colorImage testing 2019-03-06 18:02:27 +00:00
DMSrcSink.h rework colorImage testing 2019-03-06 18:02:27 +00:00