-
-
Notifications
You must be signed in to change notification settings - Fork 2
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Switch to using build
(instead of pep517
)
#233
Conversation
@duckinator I ran the tests, and it builds one of my local projects without issue. |
I'm confused why the linting fails in the Zippapp bootstrap (and only then) ; I don't even see where does the import-order linter come from, let alone whether it is erroneously treating |
The current pylint setup is to enable everything, and disable what we don't want. I'm okay adding Also: the zipapp kludge is still needed, unfortunately, because the |
@nbraud I'd say try reverting the commit removing the zipapp kludge and work from there. |
Oh sadness xD |
PS: Looks like the erroneous lint still triggers in the boopstrat tests. |
You can go ahead and add wrong-import-order to the list of disabled checkers here: https://github.com/duckinator/bork/blob/master/.pylintrc#L21-L27 I honestly thought I'd disabled it at some point. Must've been another project. 😛 |
OK, it looks like there are a few things going on:
|
c4878a8
to
5c7e01a
Compare
@duckinator Could you look into the breakage on macOS? I don't have any Apple devices to test with :3 |
@nbraud looks like we're running into pypa/build#108. Looks like the tl;dr is it's a resolved setuptools bug, and pinning setuptools>42 in pyproject.toml:
(We did similar with Emanate ages ago, for different reasons, in duckinator/emanate/pull/136) |
235: Fix CI r=nbraud a=nbraud - [x] Stop running lints when running tests. Not only is this inefficient, it forces us to get linting working in all environments, which isn't currently possible (see pylint-dev/pylint#3882). As a result, remove the `_test-only` task. - [x] Don't run tests when bootstrapping Bork in CI - It is useless, as `python3 /tmp/bork...pyz run test` immediately shells out to pytest, and tests the version of bork that is on the filesystem. - It fails because `git` isn't installed. Changes by @duckinator and myself, extracted from #233 and #234 Co-authored-by: Ellen Marie Dash <[email protected]> Co-authored-by: nicoo <[email protected]>
Using pep517 through its CLI entrypoint is not recommended. Closes: #197
Otherwise, binary wheel packages cannot be built.
This is necessary, as `build` otherwise fails on macOS.
Older versions are buggy on macOS; see pypa/build#108
Looks like version constraints in For now, I kicked in a setup step on macOS that updates |
Obsolete setuptools was the cause of a CI bug on macOS, see previous commit.
I know that pip is working on fixing this problem on their side. It was part of their big rewrite of the resolver, iirc. I'll investigate and see if I can find what exact thing is responsible for ignoring the version here. |
In theory we should be able to get away with just updating pip, which is probably good practice anyway. But yeah, since it's giving us problems, let's install new setuptools as well. |
Opened pypa/pip#9416 about being unable to pin requirement versions in |
From the discussion there, it looks like If you can reproduce the issue locally (I do not have a macOS box to test with), could you confirm that |
This reverts commit 9e3efc1.
@nbraud I do not have access to a mac, so can only reproduce it in CI. I'm not entirely sure how to approach this, tbh. |
Oh, OK. I'd suggest merging this anyhow, since the issue is shown to be outside of |
@nbraud sounds like a plan to me. merge when you think it's ready. 👍 |
bors r=duckinator |
No description provided.