Aurelia Molzer

Addendum: For extensions that do not match any known media type, translate them to an 'opaque' media type similar to opaque host that only matches that extension itself.

The builtin `ImageFormat` type is `const`-constructible and exhaustive in the sense of having an `all()` iterator. It also reflects the status of feature flags dynamically. We probably want to keep...

Performance seems to have been an artifact of the known problems llvm. Here's the results from Rust `1.89` locally: ``` test result: ok. 0 passed; 0 failed; 82 ignored; 0...

I feel uneasy about the fuzzer results. On the one hand, the kind of finding is just confirming the known bug from #605 and that's been around for a bit—we...

Note: rebased on top of #644. There's no fuzzing failures _yet_ that are detected for that PR alone but in previous iterations they did detect some here. I think there...

I was testing this locally in combination with #640 and that seems to be important. I was quite defensive with the implementation of the flush behavior and it won't indicate...

Turns out the comprehensive fix seemed _even_ harder that the above comment. It goes like this: the currently published code detects whether the IDAT sequence (or fdat) is finished by...

This seems to _improve_ performance. Some quite significantly. (I think it mostly avoids the weird forced buffer allocations we previously had in the flush call). Alas, it's almost suspicious so...

> I did pr just to share work. Even though I think we won't merge this (interop with `ExtendedColorType` is not super well defined, having an implied buffer layout visible...

I get where your need is coming from but there's a significant difference to the EXIF. For exif there's a standard / specification that describes how the _same_ semantic data...