[bug#62252] [PATCH v3 09/19] gnu: python-hatchling: Use upstream homepage.

[Liliana Marie Prikler]

Am Sonntag, dem 26.03.2023 um 09:55 +0200 schrieb Ricardo Wurmus:
> Hi,
> 
> I have built python-orjson on wip-rekado-rust-team, which is based on
> top of the rust-team (but rebased on top of current master).  This
> lets us build python-fastapi without having to remove orjson.
Do we have an ETA on wip-rekado-rust-team?  And most importantly, does
it come with antioxidant? :)

> (I worked on packaging python-fastapi for python-pytorch-lightning.)
> 
> My python-fastapi package looks like this:
> 
> --8<---------------cut here---------------start------------->8---
> +(define-public python-fastapi
> +  (package
> +    (name "python-fastapi")
> +    (version "0.95.0")
> +    (source (origin
> +              (method url-fetch)
> +              (uri (pypi-uri "fastapi" version))
> +              (sha256
> +               (base32
> +               
> "062fy794mv0yfb4ac5yaaqh6a0k7npa0n7dc4x0a5ncx1sqzvm4r"))))
> +    (build-system pyproject-build-system)
> +    (arguments
> +     (list
> +      #:test-flags '(list "tests")))
> +    (propagated-inputs (list python-pydantic python-starlette))
> +    (native-inputs (list python-anyio
> +                         python-black
> +                         python-coverage
> +                         python-databases
> +                         python-email-validator
> +                         python-flask
> +                         python-hatchling
> +                         python-httpx
> +                         python-isort
> +                         python-jose
> +                         python-multipart
> +                         python-mypy
> +                         python-orjson
> +                         python-passlib
> +                         python-peewee
> +                         python-pre-commit
> +                         python-pytest
> +                         python-pyyaml
> +                         python-sqlalchemy
> +                         python-types-orjson
> +                         python-types-ujson
> +                         python-ujson
> +                         python-uvicorn))
> …
> --8<---------------cut here---------------end--------------->8---
> 
> The update of pathspec looks good to me, though with 300+ rebuilds
> this whole thing should probably be built out on a feature branch
> first.
Hmm, 380 rebuilds would be fine if it was just python imho, but since
it appears to cut into R and Gnome as well, I get your point.  Should
we do the typical -next dance?

> The starlette update also looks good to me, though you could move it
> to 0.26.1 (with hash
> 1gd9pvcfifrlaskqz741n4rbsgljn6jnjgkscvj211paay87knj1).
> I used the version form pypi which didn’t come with tests, so I don’t
> know if running the tests of 0.26.1 is more difficult than running
> those of 0.25.
I actually haven't tested 0.26.1 – the patch was lying around for some
while before committing, since I didn't have the other parts running
back then.  I'll probably send a v4 once I check whether there's an
issue and what that issue might be.

Cheers