Skip to content
Proyectos

bradfitz/exp-httpclient

Ver en GitHub

bradfitz/exp-httpclient has 582 GitHub stars and 12 forks. GitHot measured 0 stars in 24 hours and -1 in 7 days. Last checked 2026-11-21 06:52 UTC. Last push 2019-01-01.

Cómo se miden estas cifras

Crecimiento de estrellas medido

in 24h
0
in 7 days
-1
in 15 days
Recopilando historial
in 1 month
Recopilando historial

Cambios netos en períodos que se solapan, no totales diarios. Los datos no disponibles se dejan en blanco. Las estrellas indican interés, no calidad.

Acerca de

experimental new HTTP client API for #golang (WIP)

Herramientas de desarrollo

README.md

Original

Rethinking Go's HTTP client

This repository explores redesigning the API for the Go language's net/http Client and Transport.

Initially, though, it collects problems with the current API. The actual solution has not yet been designed.

FAQ

What's wrong with Go's HTTP client?

See the list of problems.

Or see the overview presentation for the problems in a different format.

What about the Server?

This repo does not aim to address the server side of the net/http package. The server half is in better shape than the client, and it's also easier to fix the client half without fragmenting the ecosystem. Changing the Server interface needs to be done much more carefully.

But even long term, it's almost certainly best for the client and server to live in separate packages. They might share some types & code from shared HTTP package(s).

Who's leading this effort?

Brad Fitzpatrick, @bradfitz. I've owned the net/http package for over 8 years and have plenty of gripes about it. I welcome all input. If we're going to finally change it, we should get it right, so there's no need to rush this process.

Contributing

This repo is temporary and doesn't accept PRs and issues are disabled. It will move at some point to Go's repos with Go's bots and policies.

For now, discuss at https://github.com/golang/go/issues/23707

What's the plan?

Roughly:

  • Iterate on the API & godoc repeatedly until it looks right (with a fake, panic("TODO")-only implementation)
  • Discuss, revise.
  • Add a temporary implementation (likely inefficient), wrapping the existing net/http Client.
  • Port code to use it. See if we're still happy.
  • Discuss, revise.
  • Copy net/http and golang.org/x/net/http2 code into httpclient (likely several packages).
  • Benchmark, tune, revise API as needed.
  • Redo the "legacy" net/http and golang.org/x/net/http2 client APIs as wrappers around httpclient

Of course, this is all up for debate.