Respec: Weirdness of "application/x-www-form-urlencoded"

[marcoscaceres]

In xref search, "application/x-www-form-urlencoded" is categorized as a HTML attribute... might be a bug in HTML that it's being exported incorrectly?

We are suggesting it be cited as "[^application/x-www-form-urlencoded^]", which doesn't actually match anything.

[sidvishnoi]

This seems to be an issue with HTML spec. Bikeshed is treating it as:
https://github.com/tabatkins/bikeshed-data/blob/260ae9181b3940eb7301a1446c64e3eddc09b206/data/anchors/anchors-ap.data#L888-L896

[sidvishnoi]

[^application/x-www-form-urlencoded^] is buggy for sure. It'll look for "x-www-form-urlencoded" attribute for "application" element. Need to allow escaping / when it's part of the term.

[sidvishnoi]

@marcoscaceres What do you prefer for escaping? [^foo//bar^] or [^foo\/bar^] or something else? First one looks cleaner, while second one will be easier in code in future (can make use of regex lookahead assertions) There are more terms like "text/html" that need escaping.

I couldn't figure out what Bikeshed uses - maybe it's not supported there.

[marcoscaceres]

I think [^foo\/bar^] as we use \ to escape citations already.

[sidvishnoi]

Looks like your comment escaped and rendered wrong?

[marcoscaceres]

Indeed, rendered wrong... had to escape it 😅

[sidvishnoi]

This issue can be closed now I think.. until we find a valid case where we need element attr that contains / (or when we support attr-value).

[marcoscaceres]

Ok, but does the HTML spec still need to be fixed so it exports the right thing?

[sidvishnoi]

Depends on what you want to export from there. I think you were looking for some serializer, which is exported