deps: update axe-core to latest #7020
Conversation
brendankenny
requested review from
exterkamp,
patrickhulce and
paulirish
as code owners
|
addressed problems in #5565 (comment):
also fixes #5940 |
| @@ -42,7 +42,7 @@ test_script: | |||
| # FIXME: Exclude Appveyor from running `perf` smoketest until we fix the flake. | |||
| # https://github.com/GoogleChrome/lighthouse/issues/5053 | |||
| # - yarn smoke | |||
| - yarn smoke ally pwa pwa2 pwa3 dbw redirects seo offline byte metrics | |||
| - yarn smoke a11y pwa pwa2 pwa3 dbw redirects seo offline byte metrics | |||
There was a problem hiding this comment.
no idea why this was ally, but that's terrible :)
| "nodes": [ | ||
| { | ||
| "impact": "serious", | ||
| "html": "<h2>Do better web tester page</h2>", | ||
| "target": [ | ||
| "div > h2" |
There was a problem hiding this comment.
regression or good thing?
There was a problem hiding this comment.
regression or good thing?
I mean, I guess it's accurate, so that's enough? It is the only h2 in the page. I'm not sure if they recently optimized their selector selection for length or something. Agree they're worse for humans, though.
| "nodes": [ | ||
| { | ||
| "impact": "critical", | ||
| "html": "<img src=\"lighthouse-480x318.jpg\" width=\"480\" height=\"57\">", | ||
| "target": [ | ||
| "body > img[src$=\"lighthouse-480x318.jpg\"]:nth-child(5)" | ||
| "img[height=\"\\35 7\"]" |
There was a problem hiding this comment.
this definitely seems worse :(
| "snippet": "<input type=\"password\" onpaste=\"event.preventDefault();\">" | ||
| }, | ||
| { | ||
| "impact": "critical", | ||
| "html": "<input type=\"password\">", | ||
| "target": [ | ||
| "body > input[type=\"password\"]:nth-child(18)" | ||
| "input:nth-child(25)" |
There was a problem hiding this comment.
yeah a lot of these seem less descriptive
| "snippet": "<input type=\"password\">" | ||
| }, | ||
| { | ||
| "impact": "critical", | ||
| "html": "<input type=\"password\" onpaste=\"return false;\">", | ||
| "target": [ | ||
| "body > input[type=\"password\"]:nth-child(19)" | ||
| "input[onpaste=\"return\\ false\\;\"]" |
There was a problem hiding this comment.
but some are betterish.... strange
| } | ||
| }, | ||
| { | ||
| "node": { | ||
| "type": "node", | ||
| "selector": "body > img[src$=\"lighthouse-480x318.jpg\"]:nth-child(6)", | ||
| "path": "3,HTML,1,BODY,7,IMG", | ||
| "selector": "img[height=\"\\33 18\"]", |
There was a problem hiding this comment.
yeah IMO these bad boys are real regressions :/
I guess we could file an upstream issue and hope they improve it soon?
There was a problem hiding this comment.
I guess we could file an upstream issue and hope they improve it soon?
Looks like it might have been dequelabs/axe-core#731. One of the comments:
I'm also not a big fan of dropping the parent selector part. One of the selectors i saw because of that was
b. Nice and short, but it seems unlikely the user will just be able to guess where that is.
but looks like they were ok with that.
Taking over from #5565.
The PR is so big mostly because all the
Learn morelinks now point to the 3.1 docs instead of the 2.2 ones and I updated theAccessibilitysample artifact.fixes #5940