python
Remove almost all unpaired backticks
As reported in #117847 and #115366, an unpaired backtick in a docstring tends to confuse e.g. Sphinx running on subclasses of standard library objects, and the typographic style of using a backtick as an opening quote is no longer in favor. Convert almost all uses of the form
The variable `foo' should do xyz
to
The variable 'foo' should do xyz
and also fix up miscellaneous other unpaired backticks (extraneous / missing characters).
No functional change is intended here other than in human-readable docstrings, error messages, etc.
I've left ./configure and friends alone because that isn't going to impact downstream tools and feels like a lot of churn.
📚 Documentation preview 📚: https://cpython-previews--119231.org.readthedocs.build/
I think you need to adjust this (or restore the error message):
https://github.com/python/cpython/blob/c0d81b256604a1079349d82d136db43eefcb3df1/Lib/test/test_enum.py#L447-L451
@geofft To me this PR now looks much bigger than necessary. This quoting style may be uncommon, but in most contexts there's nothing particularly wrong with it. I think the Sphinx failures are sufficient reason to avoid it in docstrings that may end up being parsed by Sphinx, but I don't see much reason to cause churn in other files (e.g. the docs Makefile, or various code comments, or especially in error messages which could in principle affect user code.) So I'd personally feel more comfortable limiting this change to docstrings.
Also, rebasing on latest main might fix the docs failure in CI.
I'd personally feel more comfortable limiting this change to docstrings.
OK, let me split that out, thanks for the review!
Also, rebasing on latest main might fix the docs failure in CI.
Yes, it will. I raced with a change to the CI rules.
OK, I think this is now just docstrings + a couple of comments in files where I was already making docstring changes.
Could you update the generated files?
diff --git a/Misc/sbom.spdx.json b/Misc/sbom.spdx.json
index 52fc058..b60adcf 100644
--- a/Misc/sbom.spdx.json
+++ b/Misc/sbom.spdx.json
@@ -188,11 +188,11 @@
"checksums": [
{
"algorithm": "SHA1",
- "checksumValue": "be21968aa6e358cb2f193f96d05df806fdf1575c"
+ "checksumValue": "fed1311be8577491b7f63085a27014eabf2caec8"
},
{
"algorithm": "SHA256",
- "checksumValue": "debff6d64b3ef4915873857c315a43193e3fa3e51be11ea19bcbc9d0451985dd"
+ "checksumValue": "3dc233eca5fa1bb7387c503f8a12d840707e4374b229e05d5657db9645725040"
}
],
"fileName": "Modules/expat/xmlparse.c"
This looks good to me. There are changes to Misc/sbom.spdx.json but the CI check doesn't seem happy with those changes. Does it change anything if you run make regen-sbom again?
From the https://devguide.python.org/getting-started/pull-request-lifecycle/#quick-guide
In order to keep the commit history intact, please avoid squashing or amending history and then force-pushing to the PR. Reviewers often want to look at individual commits.
Oh, I didn't realize CPython does squash-and-merge. Thanks for the pointer.
In this case you can use the "compare" link on the right to compare the diff - the change from the previous is just Misc/sbom.spdx.json.
That wouldn't work with multiple commits, it's easier to look through the history from the commits tab and you get a useful overview off all changes since you last looked in notifications.
@geofft, please do not force push, use git push or git merge <branch>.
Show schema
Note: if you use squash, it doesn't commit.
@nineteendo, thanks for helping review PRs by other contributors and letting them know about our usual policies and workflow. However, please try to avoid using ALL CAPS when addressing other people in the CPython repo. While I'm sure you didn't mean it in this way, if someone uses ALL CAPS online, it can give the impression that they're shouting or angry :-)
@nineteendo, I was rebasing to resolve a merge conflict. Is there really a preference to resolve merge conflicts by adding merge commits? Another reviewer above specifically requested rebasing (not merging).
My reading of the document you linked was that avoiding force-pushing was a preference, not a requirement. In this case, adding a merge commit would have made it harder to figure out what I did, I think?
In any case - I think this is ready to merge now, can we merge it before there are further merge conflicts? :)
In this case, adding a merge commit would have made it harder to figure out what I did, I think?
Actually not, because GitHub shows a condensed view in that case instead of all unrelated changes. I was already wondering if you made other changes.
In any case, if you force push, always explain why you did it and why you didn't make a merge commit.
Merged, thanks @geofft for the contribution!
(My fault on confusing the issue re merge vs rebase. It is true that typically even for resolving merge conflicts we do merge rather than rebase and force-push, but sometimes we do still casually use the word "rebase" for this 🤦 But none of this is a hard-and-fast rule, there was no problem with your force-push on this PR. The main reason to prefer merge is to avoid accidentally invalidating someone's in-progress review comments.)