Flush support

[fasterthanlime]

When using brotli as transport compression, it's useful to be able to force flush between messages.

edit: compression, not encryption 🐙

[fasterthanlime]

I implemented this in itchio@529f981 but my initial tests make me think it's unreliable (although there were a few other things in my pipeline, so a properly-isolated test would be better..)

[kothar]

I've merged your Flush method in #27 @fasterthanlime, but rearranged the encoder slightly to factor out the flush behaviour between Write, Flush and Close.

Unfortunately the tests fail - can you take a look to see if I've done something stupid? Decoding a buffer if you've used Flush seems to fail - is that what you've seen in your tests?

[fasterthanlime]

@kothar just looked, not seeing anything stupid in #27, and yes, that's the behavior I was experiencing.

Either there's something fishy upstream, or we're not using the API the way they thought users would?

[fasterthanlime]

Maybe @dsnet has some insight on this? 👯

[aaronfranke]

@fasterthanlime has forked this repo and updated it, perhaps it's up to him now.

https://github.com/itchio/go-brotli

[fasterthanlime]

I haven't told @kothar about my fork yet :) Let's do it now:

I wanted to update to the latest upstream, and

• some features were deleted (custom dictionary support - I hear it's coming back soon)
• I also changed the interface a little bit (removed BrotliParams, it's now a simple struct)
• the Compress/DecompressBuffer tend to not take an input buffer in my version - decompressed size estimation code is gone too
• the Writer in particular has changed a lot, it's heavily inspired by the cbrotli package that upstreams ships. Their encoder is now C-only, so it has a clean interface and we don't need encode_go.cc anymore
• my fork isn't using gopkg.in - it's vendored in the projects using it anyway

I didn't want to be a bother so it's in a fork for now!

[kothar]

Cool, I'd not made much progress updating to the more recent upstream changes, so thanks for keeping on top of it! Happy to merge at some point, but sounds like it would be a breaking change to the existing API, so we should consider bumping the release version if we go that route. Also fine if you're happy to keep maintaining your fork and direct people there. Mike.

…

On Thu, 11 Jan 2018, 2:51 pm Amos Wenger, ***@***.***> wrote: I haven't told @kothar <https://github.com/kothar> about my fork yet :) Let's do it now: I wanted to update to the latest upstream, and - some features were deleted (custom dictionary support - I hear it's coming back soon) - I also changed the interface a little bit (removed BrotliParams, it's now a simple struct) - the Compress/DecompressBuffer tend to not take an input buffer in my version - decompressed size estimation code is gone too - the Writer in particular has changed a lot, it's heavily inspired by the cbrotli package that upstreams ships. Their encoder is now C-only, so it has a clean interface and we don't need encode_go.cc anymore - my fork isn't using gopkg.in - it's vendored in the projects using it anyway I didn't want to be a bother so it's in a fork for now! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#25 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAXFPs2pNFFWN5buzycx3F1K9K4g-dr-ks5tJh_2gaJpZM4Gmnop> .

[kothar]

I have added a link to https://github.com/itchio/go-brotli to the project description, and as I'm unlikely to be working on this binding in the immediate future I'll archive this repo. I'll submit any future changes to your repo @fasterthanlime 👍