Re: [PATCH 0/1] Fix netrc usage with WebDAV-based HTTP protocol

From: brian m. carlson <sandals@crustytoothpaste.net>
Date: Feb 24, 2025, 22:09 UTC
Message-ID: <Z7zuMBVrnX_pLE0K@tapette.crustytoothpaste.net>
In-Reply-To: <xmqqa5ab70ka.fsf@gitster.g>

On 2025-02-24 at 16:46:29, Junio C Hamano wrote:

Show 10 quoted lines

> "brian m. carlson" <sandals@crustytoothpaste.net> writes:
> 
> > However, I think we need to either decide that we're going to support
> > this as a fully-fledged feature and add documentation and sufficient
> > tests, or remove it, since it's not reasonable to have undocumented,
> > untested features that people rely on.
> 
> Does "this" refer to "use of netrc in HTTP code paths", or "HTTP
> walker"?  I am kind of surprised that anybody is still using the
> latter, regardless of where the credential is locally stored.

It refers to netrc in HTTP code paths.

There are definitely people using the WebDAV-based HTTP protocol still, but after reading the comment in the code that implies that we don't read packed-refs, I'm less and less inclined to say that's a good idea. It's probably going to die pretty quickly when we move to reftable, though, since that protocol only works with the files backend.

-- 
brian m. carlson (they/them or he/him)
Toronto, Ontario, CA