-
Notifications
You must be signed in to change notification settings - Fork 321
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
MAINT: Use different lighthouse action #548
Conversation
It seems a lot of accessibility stuff was introduced by @bollwyvl here: #294. Nick, WDYT about the py11y removal proposed by @choldgraf here? |
Btw, @choldgraf, it seems there is some branch conflict. |
Yes I would love to hear @bollwyvl's thoughts re: maintainability+complexity vs. doing a reasonable job of addressing accessibility issues. I think the main question I have is: what are the benefits to using |
i think it's a regression in development and testing robustness if there is no local capability for a developer to identify (and therefore fix) the ~30% of WCAG issues that To that end, I don't even know what claims lighthouse (or the wrapper action) are making, as it's more of a general-purpose tool, and was useful for putting number to things like asset pre-loading that helped us improve the perceptual speed of theme with actual metrics, and helping keep it there (provided we haven't just been bumping the thresholds down). The fact that it handles some accessibility concerns seems almost incidental, aside from google's goal of helping you convert more eyeballs (or screen readers) so they can sell more ads and personal data. Longer term, I'd prefer that concerns like accessbility aren't delegated to a secondary test suite: i'm also looking into working accessibility testing into more real browser environments, with e.g. robotframework/selenium, but don't yet have anything concrete to demonstrate these. The advantage there is one could hijack things like But same deal applies: if a tool can't be run locally and in CI worker, I feel like it's bad for open source. |
Just a note that you can generate lighthouse reports directly from Google Chrome. So at least that's one option for local running, but as you say the things that lighthouse vs. pa11y test for are slightly different. I agree with all of your points, I just worry that right now our usage of pa11y is not super actionable. As a first step, I've added some more documentation of what the pa11y error messages mean in the contributing docs: https://pydata-sphinx-theme.readthedocs.io/en/latest/contribute/topics.html#errors-in-ci-cd-and-what-to-do Maybe this will make it easier for people to understand what the pa11y test does and how to interpret the results? |
Two more updates on the pa11y stuff:
|
I think that the accessibility audits with pa11y was just flaky, not broken. I also opened up a new PR to reduce the scope to just lighthouse, and not touching pa11y. closing this one! |
This uses a dedicated GitHub Action for lighthouse to analyze and upload reports for our HTML artifacts automatically. It simplifies the configuration of our pages, and also provides reports that are a bit more understandable than our current setup.
It removes the
pa11y
as well - my reasoning for this is the following:Question: remove
pa11y
?There's a lot of other scripting stuff that we could remove if we want to stop relying on pa11y for our accessibility. I think there are pros and cons, but wanted to show off this alternative as a start. What do others think?
closes #546