Conversation
florimondmanca
left a comment
florimondmanca
left a comment
There was a problem hiding this comment.
Changelog looking good 👍
|
|
||
| ### Added | ||
|
|
||
| * Add support for `Client(allow_redirect=<bool>)`. (Pull #1790) |
There was a problem hiding this comment.
Typo: Is it allow_redirectS?
There was a problem hiding this comment.
It is indeed!
Or follow_redirects with #1808. 😬
…to update-changelog
|
So... #1808 is probably way to big of a behavioural change to rush into. I figure we may as well get this out the door now as 0.19.0, and just have a little more time to mull the last bits of how do we really, really want the 1.0 API to look?. |
|
|
||
| The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/). | ||
|
|
||
| ## 0.19.0 (19th June, 2021) |
There was a problem hiding this comment.
@tomchristie should this be August 18th/19th (not sure your timezone)? I'm slightly surprised to see something mentiong a couple months ago, but it's been released ~9hours ago.
There was a problem hiding this comment.
Oopsie — looks like that got merged without a change?
There was a problem hiding this comment.
Thanks for updating, I believe the GitHub release title can also be updated here https://github.com/encode/httpx/releases/tag/0.19.0
3 participants
So... I guess here's the lowdown.
Final set of changes, that we'd release in a 0.19.0 version, and then promote to a 1.0 in perhaps 2 weeks time. (I'm away on holiday this following week 23rd-26th, so getting these changes out a few days prior to that would seem to be a good idea.)
Also requires Switch follow redirects default #1808Might want to document the redirect differences from requests more prominently?__version__.Regarding 1.0... I'm not 100% convinced that the transport API is necessarily absolutely definitively the most favoured option, but it seems silly at this point to keep blocking everything indefinitely. It's a little verbose, but it is a nice cut-off between the two slices of functionality.
There's a really nice roadmap ahead, and we should probably call it at this point, so that we can push on with:
httpcoreimplementation that starts to really neatly expose some lower-level connection pooling details, as well as fixing a couple of gnarly implementation issues we currently have.httpxcommand-line client, which would just be a fantastic piece of functionality to offer, for very, very little additional code.