Missing license text for vendored httpx-ws

[musicinmybrain]

Since #1042, httpx2 includes httpx2.websockets, which is derived from (and is largely a vendored copy of) httpx-ws, https://github.com/frankie567/httpx-ws. This is reasonably well documented in the docstring,

httpx2/src/httpx2/httpx2/websockets/__init__.py

Lines 2 to 4 in cad5353

	WebSocket support, derived from httpx-ws (https://github.com/frankie567/httpx-ws).
	Copyright (c) 2021 François Voron, MIT License (https://github.com/frankie567/httpx-ws/blob/main/LICENSE).

but the MIT license says,

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

so a mere link to the license text does not suffice; a copy of https://github.com/frankie567/httpx-ws/raw/refs/tags/v0.9.0/LICENSE should be included somewhere in the source of httpx2.

(I noticed this while updating the python-httpx2 package in Fedora.)

I would be happy to open a PR for this, but I’m not sure where you would prefer to put the text.

• Inside the vendored part of the package, src/httpx2/httpx2/websockets/LICENSE?
• Appended to the “main” license file LICENSE.md, with a brief description of which portion of the code it applies to?
• In a file like LICENSE-httpx-ws, src/httpx2/LICENSE-httpx-ws, etc.?
• Something else?

[Kludex]

src/httpx2/httpx2/websockets/LICENSE please 🙏

I'd appreciate if you could do the same for SSE 🙏

[musicinmybrain]

PR proposed in #1087. Thanks for the input!

One more related question while we’re talking about this: would you consider adjusting

httpx2/src/httpx2/pyproject.toml

Line 17 in cad5353

license = "BSD-3-Clause"

to BSD-3-Clause AND MIT to reflect the fact that httpx2 contains code under both licenses?

[Kludex]

I would prefer to avoid that - Do you have real examples that did that?

[musicinmybrain]

I would prefer to avoid that - Do you have real examples that did that?

I could probably dig up a couple if I tried, but I could also dig up a couple that declined to.

The basic question is whether this field should document the “main” license, or whether it should cover all of the licenses that apply, including bits of code copied from other places. Until the relatively recent shift to SPDX expressions with PEP 639, the latter approach wasn’t really feasible in a clear and unambiguous way. For RPM license metadata in Fedora, we started with an “effective license” theory and moved firmly to full enumeration of all licenses as we transitioned to SPDX, but upstream Python license metadata doesn’t have to be aligned with our way of doing things.

In the Python ecosystem, I haven’t observed a real consensus on how this field should be formed. The examples at https://packaging.python.org/en/latest/specifications/license-expression/, including things like MIT AND (Apache-2.0 OR BSD-2-Clause) and MIT OR GPL-2.0-or-later OR (FSFUL AND BSD-2-Clause), show that comprehensive license enumeration was anticipated, but that doesn’t mean that there is a culture of actually doing things that way. This is in contrast to Rust, where license metadata has been SPDX expressions from the beginning, and I’ve found that upstreams are mostly happy with PR’s that add terms to their license expressions.

I stand by the suggestion, but only as a suggestion. It doesn’t cause me any great pain if you decline.

[Kludex]

Then I'd prefer to decline it. If someone is offended about this, we can add it later.

[Kludex]

Thanks for the context.

[musicinmybrain]

Fixed by #1087. Thanks!