Current priorities

Some thoughts on general priorities right now.

There’s several projects that I think we can working towards a 1.0 release…

http3

(Hopefully released as requests3 at some point)

  • Addressing any reliability issues.
  • Using http3 for Starlette’s TestClient rather than requests. (It’ll allow us to provide both a sync and an async variant.)
  • Documenting http3 as an option for an async HTTP client in Starlette.
  • Either documenting how to write a Proxy with http3 or including a Proxy component in Starlette.

databases

  • Locking to ensure sequential accesses on any single connection.
  • Tidying up the Records API and making it more strictly consistent across different drivers.
  • A proper documentation site.

uvicorn

  • HTTP/2 support.
  • Addressing any outstanding WebSocket issues.
  • Get project to 100% coverage and type annotation, like the others.

starlette

Cross project

  • Interlinking project docs, github pages, and discussion groups from https://www.encode.io
  • Different colour themes for each project. Make these match up with the category colour markers on this discussion site.

I also really need to spend some time getting the Django REST framework 3.10 release done.

Projects that are slighty on the backburner until other bits are nicely shored up are typesystem ,orm, and apistar. I want to come back to spending time on all of those too,

2 Likes

Another http3 bit of work that could be pretty worthwhile - releasing an httpie transport plugin, to add HTTP/2 support to httpie.

See also https://github.com/jakubroztocil/httpie/issues/692

:+1:

(Hopefully released as requests3 at some point)

I thought I read somewhere that http3 departed from Requests for good, can’t find the original post though. Did I read wrong or has the strategy changed?

Different colour themes for each project. Make these match up with the category colour markers on this discussion site.

I’d be glad to help with this! But maybe setting up the docs site for Databases (at least w/ the current documentation) is a prerequisite?

You probably mean this…

There’s been some issues around the requests’ project management, esp. transparency around the fundraiser. If we’re able to deal with those properly then it’s feasible that the project could still end up becoming the requests3 package.

I generally think that’d be in the community best interest, but it’d need to come attached with a good overhaul of eg. project funding & maintainance.

But maybe setting up the docs site for Databases (at least w/ the current documentation) is a prerequisite?

I guess so, yup.

1 Like

Have you been in contact with the PSF to see if there are any funds available for this?

I haven’t, nope. I don’t think they’ve got a huge amount of resources available ATM, and my time is just about covered by Encode’s existing model, so I’ve got that side covered.

Ideally I’d like to be able to launch this with a bit of fanfan, and run a similar funding model to REST framework’s sponsorships. I’d hope given the position of requests in the Python landscape, that there’d be enough funding potential there to take on at least a part-time requests3 maintainership/intern position.

I’ve not figured out exactly if that works okay with all other interested parties, but that’s the sort of thing I’d like to aim towards.

1 Like