Closed fergald closed 6 years ago
Can you comment more on what the actual issue is? This is just the <meta name=abstract>
element, not the actual visible abstract. It just takes the literal text that you specified in the Abstract
metadata and spams it in; any markup you use in your Abstract is similarly spammed in there.
If you wanted it to match the final output HTML that showed up in the abstract, I could potentially do that, but it would be a dramatically different architecture. What problems is this causing in pr-preview?
bikeshed spec w3c/csswg-drafts/css-shadow-parts-1/Overview.bs
produces the following
<meta content="This specification defines the ''::part()'' and ''::theme()'' pseudo-elements on <a>shadow hosts</a>, allowing <a>shadow hosts</a> to selectively expose chosen elements from their <a>shadow tree</a> to the outside page for styling purposes." name="abstract">
HTML differ chokes on the unescaped "<" in the attribute string
E.g.
Huh, that's a bug in htmldiff. There's nothing wrong with a < in attribute values.
Actually it seems like html tidy is the problem
Not the differ itself (running htmldiff.pl on the file seems to be no problem). I'm not sure how that service is set up. I'll try bug the listed maintainer.
That said, empty tags in the metadata don't seem useful.
Woah, huh, where's it got that doctype from? It might legit be wrong in xhtml, but this should all be html5 doctype
I've emailed htmldiff and htmltidy maintainers so see if we can get this fixed up.
htmldiff has been fixed!
:+1:
E.g. in css-shadow-parts-1 the following input
produces the following output
Which has several empty <a> tags.
Where this hurts is that the automatic diff tool pr-preview (which works fine if I remove the a-tags from the abstract).
Spec.fixText seems to be where the action is here but I don't know enough about it to hazard a fix but I guess the desired output is something like