This change results in losing the alpha discovery, which means we won't be able to notify Skia if the decoded bitmap is opaque or transparent. Not sure how important this is. Adding brianosman@ and mtklein@ to answer this question.

Our thinking is, most of the opacity information we need will come from the headers / image format, so it's probably fine to just drop this part of the code.

If we find it turns out to be important, I'd be happy to write up a second pass that does per-pixel alpha checks.

Blink relies on opacity checking for optimizing painting when we know things are obscured (and cc does similar things). Having said that, for decoded images we never get it right anyway because we do not update the opacity status after decoding.

So I think this will not result in a change in behavior.

Won't fix. According to the discussion with reed@, mtklein@, and brianosman@, the better thing to do is to retry replacing the current Chromium decoders with SkCodec decoders.