loona-h2-v0.2.1

Fixed

• Fix rustdoc errors

loona-h2-v0.2.0

Other

• updated the following local packages: fluke-buffet

httpwg-v0.2.1

Added

• Generate catalog in httpwg-macros

Fixed

• Fix rustdoc errors

Other

• Improve httpwg-cli, make it not so noisy

httpwg-v0.2.0

Other

• updated the following local packages: fluke-buffet, fluke-hpack

httpwg-macros-v0.2.1

Added

• Generate catalog in httpwg-macros

Fixed

• Fix rustdoc errors

httpwg-cli-v0.2.1

Added

• httpwg-cli: Count tests passing, fail when not all tests pass
• Generate catalog in httpwg-macros

Other

• Improve httpwg-cli, make it not so noisy

buffet-v0.2.1

Fixed

• Fix rustdoc errors
• Fix tls example

Other

• Avoid unused warning

buffet-v0.2.0

Added

• Upgrade dependencies

Other

• Fix tests for Linux
• Make rfc9113 pass, still need to check codes
• progress
• Flesh out the handshake test
• Migrate more tests to pipe, deprecate ChanRead/ChanWrite
• Migrate ChanRead uses over to pipe
• Adjust shutdown implementation for TcpWriteHalf (io-uring codepath)
• Port test to ChanRead
• More assertions
• Finish up pipe implementation
• Simplify WriteOwned trait, introduce pipe
• Cancel listen
• Some cancellation?
• Remove listen cancellation
• Remove WrappedOp
• wip cancellation
• Only import io-uring on Linux, try closing fds on TcpStream drop
• make io-uring dep linux-only
• All tests pass!
• Some tests are starting to almost pass with io-uring-async
• flesh out TcpListener API
• We can successfully accept a connection!
• A little print debugging
• Use nix for errno stuff
• Vendor io-uring-async and upgrade it to io-uring 0.6.x
• io-uring-async (cf. #154)
• TcpWriteHalf fixes
• Fix buffer bug
• Remove BufOrSlice, closes #153
• Bring 'fluke-maybe-uring' back into 'fluke-buffet'
• Fix linux build
• Headers flow control wip
• wip header flow control
• wip Piece
• Finish propagating Piece/PieceCore split changes
• Piece::Slice
• drat, partial writes.
• hapsoc => fluke
• Fix http2/4.2/1
• Fix h2spec http2/6.5/3
• mh
• no continuation state
• Upgrade more deps
• Upgrade memmap2
• Upgrade pretty-hex
• Upgrade deps
• Switch to Rust stable, closes #128
• Bump some dependencies
• Bump dependencies
• Opt out of async_fn_in_trait warnings
• release

httpwg-v0.1.1

Other

• updated the following local packages: fluke-buffet, fluke-hpack

fluke-v0.1.1

Added

• Upgrade dependencies

Fixed

• fix more tests
• fix several low-hanging bugs: re io/read tasks

Other

• Try out pre-commit hook
• More testing facilities
• Make rfc9113 pass, still need to check codes
• Alright, back to writing tests!
• Introduce httpwg-macros, httpwg-gen
• Woo, namespacing tests, getting rid of the useless struct. Maybe we can codegen the rest actually.
• Woahey, 4.2 test works?
• More test facilities
• Flesh out the handshake test
• 4.1 test work
• Well, well, almost got a working 4.1 test
• Introduce 4.1 test (needs read loop in the background)
• More test harness generation
• Generate tests
• Introduce start_server function
• Use pipe for httpwg tests
• Migrate ChanRead uses over to pipe
• Finish up pipe implementation
• Simplify WriteOwned trait, introduce pipe
• First test is doing things
• Introduce fluke-h2-parse
• Some tests are starting to almost pass with io-uring-async
• Fix buffer bug
• Remove BufOrSlice, closes #153
• Bring 'fluke-maybe-uring' back into 'fluke-buffet'
• Do a bit of flow control on incoming streams
• Improve code coverage, closes #151
• Fix linux build
• Implement final flow control touches
• remove unused methods
• 6.9.1 all three cases pass
• well I think 6.9.1 should pass.
• Headers flow control wip
• wip header flow control
• do flow control accounting for headers as well
• curl tests work! partial writes work.
• wip Piece
• Finish propagating Piece/PieceCore split changes
• Piece::Slice
• drat, partial writes.
• Ready to write
• a little forgetfulness
• BodyEnd sets eof to true
• Remove erroneous/outdated comment (the Frame struct no longer contains the payload)
• Pass h2spec 6.9 tests apparently, even though we're never sending data now?
• Introduce StreamIncoming, StreamOutgoing
• Start implementing flow-control
• hapsoc => fluke
• Fix h2spec http2/6.4/3
• Fix h2spec/6.5.2/1
• Fix h2spec http2/6.5/3
• More h2spec cases
• Close connection explicitly in h2spec
• Upgrade tokio-uring
• mhh
• Fix more cases
• well every time I lose silly time to this I improve debug logging, so.
• mh
• Uhm
• better debug implementation for Frame
• Remove read module, move it back into server.rs
• Remove write module altogether
• Move all writing out of write
• Retire H2ConnEvent
• mhmh
• Move settings acknowledgement inside read
• Deprecate H2ConnEvent::RstStream
• Move goaway writing away from write
• Move ping to write_frame
• Introduce write_frame
• Give out_scratch to h2readcontext
• Get rid of write task
• Fix some flow-control cases
• Fix PING tests
• Fix more tests
• Oh joy, h2spec is wrong
• More test fixes
• Fix more stream state cases
• more stream state things
• Fix some stream state cases
• Rename StreamStage to StreamState
• Implement RstStream
• Fix hpack tests
• looking good
• No more compile errors woo
• Simplifying headers/trailing reading + being more rigorous with conneciton errors
• no continuation state
• Upgrade more deps
• Upgrade pretty-hex
• Upgrade deps
• Graceful GOAWAY handling
• Set minimum rust version for crates with async fn in trait
• Bump some dependencies
• Remove TAIT
• Bump dependencies
• Opt out of async_fn_in_trait warnings
• Move fluke-h2spec somewhere else
• Move curl tests into their own crate
• Simplify H2Result code further, try sccache.exe on Windows
• Have a few methods return H2Result, which clarifies the control flow
• Fix http2/5.1/5
• Introduce 'write' module
• Move types around
• Bring state names closer to RFC
• First stab at 5.1.2
• todo => debug, don't crash on window_update for known streams
• Don't send GO_AWAY when receiving a connection-wide window update
• Send GO_AWAY if we get a stray WINDOW_UPDATE
• Extract process_frame method out of H2ReadContext::work
• release