On Wed, 14 Oct 2026 19:12:11 GMT, Sergey Bylokhov <[email protected]> wrote:
> > That does not make the plugin non-baseline. > > It definitively cannot support anything other than 8 bit samples and was > > never intended to. I know this for a fact. > > How is "8 bit samples only" related to "baseline only"? That is not the same > thing. Baseline in the spec is 8 bit samples + sequential + Huffman + 2+2 > tables. Progressive with 8 bit samples is an extended process, and our plugin > writes it (JPEGImageWriter -> native writeImage -> jpeg_simple_progression). > So the plugin is 8 bit samples only, but not baseline only. That doesn't change quantization so it is the same thing. > > And 8 bit samples do not mean 8 bit quantization tables, the libjpeg does not > enforce that: it writes and reads 16 bit tables for 8 bit samples, as the > test above shows. So "compiled for 8 bit samples" does not limit the tables, > this patch does (or tries to). Yes it does. I referenced the exact place in the JPEG spec that requires it. any use of 16 bit quantization with 8 bit samples is a bug in what ever provides it or whatever allows it. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/33217#discussion_r4275737090
