Paolo Barbolini

Another attempt to refactor this crate. Like #2 there's a ~7% performance slowdown which I can't explain. Closes #2

Multistream decoding support

2

Support decoding multistream bzip2 files. Originally posted by @ygyzys at https://github.com/alexcrichton/bzip2-rs/issues/75

This is one of those changes which should have been easy. State shouldn't live inside `Block` but inside `Decoder`. However this change seems to create a decompression speed regression up...

Currently compiling on Windows (at least on CI) fails. This is not a priority for me, since I'm not using this on Windows. I also don't have a Windows machine,...

Different attempt at #810

3

Also cleans up the encoder a lot, removing some complicated logic introduced by the initial round of implementation

Use cargo weak dependency features to reduce tokio features

2

Would unfortunately require us to bump the MSRV to 1.60, which is too high https://lib.rs/stats#rustc

With #763 and #765 reducing our reliance on the regex crate I think it's now possible for it to be removed entirely at some point. This removes the need to...

Rename `timeout` to `connection_timeout`

3

Renames `timeout` to `connection_timeout`, since that's all it's really controlling. If we were to add timeouts on more things in the future they would better work as separate parameters.

Starts work on option 4 of #693. Still very WIP and incomplete.

How to fix headers

1

#626, #672 and #690 have shown how hard it is to do headers in email. #685 tries to fix the first issue by correctly encoding the `Content-Disposition` header. That PR...