Repository navigation
Conversation
|
This can be tested with If using jj, I recommend this fileset for review: jj diff '~**/test_template.tera & ~**/Cargo.lock'This makes the diff much smaller to review. The breaking changes in the test templates aren't all that interesting (IMO), as long as the output remains identical. |
|
@iHiD a little heads up, the CI check says that translations are missing. But these test templates shouldn't be translated anyway, right? They are not user-facing. Maybe the CI job can be made more precise, to avoid unnecessary, expensive translation runs. |
|
@senekor You need to add the |
|
@iHiD Why not fix the check instead, such that the label is only required when there are actually files to translate? Track maintainers contribute during their spare time. Asking them to spend that time ticking unnecessary boxes isn't great. |
I do. There's hundreds of lines of codes that check all these things. This one got missed as it's a markdown file in a directory where nearly all the markdown files do get translated. I'll add more checks that check that this file gets rendered on the website.
I made a mistake by missing this edge-case. I'm sorry it upset you, but I think this reaction and comment is a little unjustified. Bugs happen sometimes, and I invest a lot of time and energy into making things like this as easy process for maintainers as I can. |
The track-docs pattern matched every docs/*.md in a track, so a PR that touched docs/CONTRIBUTING.md failed i18n / completeness and asked for a translation nobody reads (exercism/rust#2187). The website shows only docs/ABOUT.md and the paths listed in docs/config.json. translatableFiles now reads docs/config.json at the ref being processed, next to the wip rule. With --base, a doc that the list starts naming is required even if its file is unchanged. A missing or unparseable docs/config.json requires every doc, as an unreadable config.json marks nothing as wip. english-changes.mjs reads the same list through its blob fetch, so the queue agrees with the check. Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
exercism/i18n#459 adds the extra check and this PR can now be merged. |
|
Thanks, Jeremy.
I don't mind bugs or mistakes, not at all. I make plenty of them myself 🙂 The sentence: "You need to ..." is what rubbed me the wrong way. To me, it read like an instruction to do a nonsensical thing. Exercism is not the only project I contribute to, and many of them are currently changing due to LLMs. I'm weary of places I love turning into ones I don't recognize. Please excuse the misunderstanding. |
|
Thank you. In the week you posted this, I probably had a dozen different people message asking how to get their PRs over the line because they didn't read the post about the labelling existing. I just missed in this thread that you'd read about it and the label wasn't needed for this PR, and responded trying to be helpful to the more general cause. Genuinely sorry that I didn't read your message more carefully.
Yeah - I get this. I think people are feeling this a lot atm, and it's hard to work out where the line between maximising LLM opportunity and making things rubbish sits. For me, I want to use LLMs to serve our students better, but it also risks making places less fun for the community (about like the whole path of development in general atm). It's tricky!
Ditto :) |
1d16ad9 to
1582ce9
Compare
No description provided.