Lukasz Anforowicz

One way to avoid this crash is to modify `rustfmt/src/ignore_path.rs` as follows: * Add a call to `canonicalize` in `IgnorePathSet::from_ignore_list` (error handling seems ok to me): ``` impl IgnorePathSet {...

### Testing I am not sure why this problem is not caught by `cargo test` in `rustfmt`. I am guessing that `rustfmt\src\test\mod.rs` always operates on canonicalized paths and therefore avoids...

It seems that paths of files being formatted are already being canonicalized in [`src/bin/main.rs`, in `fn determine_operation` / when building `Operation::Format::files`](https://github.com/rust-lang/rustfmt/blob/40f507526993651ad3b92eda89d5b1cebd0ed374/src/bin/main.rs#L505). And I've tested that therefore it is sufficient to...

Good catch. I think option 1 is undesirable - I also think that the current content seems incorrect - AFAIU `Box::new(...non-ZST...)` will always return a non-null pointer (and this means...

Any updates? :-) (One other breaking change that I'd like to see merged is https://github.com/image-rs/image-png/pull/421 - maybe it's time to pull the trigger on both of them? I haven't looked...

Thanks @Shnatsel, @kornelski, @okaneco, and @fintelia for the PR, reviews, and for merging! Any chance that you could please push a new release/version to https://crates.io/crates/png/versions? A new official version would...

Ooops... I can repro the fuzzing problem locally. Let me convert this PR into a "draft" for now. I'll ping if/when I am able to fix this. Sorry about the...

Let me explain the problem that was caught by `buf_independent` fuzzer: * The suffix of an `IDAT` payload contains some bytes that are *not* needed to decompress the image data....

In the force push above I am tweaking this PR one more time to explain this errors-ignoring-behavior in a doc comment of `next_frame_info` ([here](https://github.com/image-rs/image-png/pull/533/commits/84b2d3789ad632b578bd284bb27b05d3a40f3445#diff-b33eb71617447baaf7837be5df8b23110737ddc23b0a37736d552bf782b4c6eeR328-R333)). PTAL. I admit that this behavior...

Is it fair to say that the buffer inside `BufReader` is replaced with the `idat_stream: Vec`?