Improve the styling of the conformance results page#2307
Conversation
- `<span>` cannot be a parent of `<p>` - `#bottom` must be unique
Add `th[scope]` attributes and split groups into separate `<tbody>`s.
Partial compliance is given half a mark, a pass a full mark.
To reduce the noise when reviewing, hide the generated output of the conformance test results.
a05617c to
99f1536
Compare
|
I timed that terribly badly... 😂 Rebased on top of #2308. You can preview the changes here. |
Are you really honest here? I would not have guessed that you wrote the issue description in a hundred years. This looks 100% like AI. |
Urgh. Yes. I 100% wrote it. I lament this day and age where everything could be considered AI or look like AI. I've frequently used em-dashes in the past and now that's considered a "tell" for AI. So tired of it 🙁 It's disappointing that you have to start with a presumption of dishonesty. |
This reverts commit 99f1536. Turns out this wasn't a good idea and broke testing in CI.
|
I'm sorry if you actually wrote it. But in 15 years of open source I've never seen this style. I have never seen anybody use emojis like this (except LLMs) and when looking at the last ~10 pull request of yours and a couple random ones in the last ten years I don't see a single Emoji. So I feel like my intuition wasn't really that weird. But as a long-term contributor to the Python ecosystem I'm obviously willing to give you the benefit of the doubt. LLMs aren't banned on this repository, and there is no necessity to disclose this although very obviously happy to have this information. In general I like the PR, I've glanced over it and there are a lot of good changes in there. I think somebody else needs to review the visual aspects, since I'm not really good at that part. |
I've been working in a private repository that has a template with headers like that and I guess I just had it in my head. Have reverted back to my usual more plain style. |
Issues Identified
Looking at the conformance results page, there were a number of issues that I noticed:
idattributes,<span>parent of<p><th>used for all table cells, not only header cellsclassattributespointercursor, even if there is no tooltip (or other action)List of Changes
This pull request aims to improve the generated output and make it easier to maintain.
The following changes have been made:
tomllibfrom the standard library instead oftomliconformance/pyproject.tomlhasrequires-python = "==3.12.*"jinja2for templates instead of manually joining lists of tags<th>for heading cells, not all table cells<thead>for the header row with type checker names<tbody>for each group of testsscopeattributes for header cells<colgroup>and<col>to reduce excessclassattributesclassattributeshelpcursor instead ofpointer, and only where there are tooltipslight-dark(), etc.1for a pass and0.5for a partial resultPasses validation on validator.w3.org (with the exception of a known bug in the validator for nested CSS) and also scores 100 for performance, accessibility, and SEO in Google Chrome DevTools Lighthouse tool.
Future Ideas
AI Disclosure
No AI was involved in creating this pull request.
Screenshots
First Load
Sticky Table Header
Tooltip Rendering
Theme Selector