"textarea wrapping transformation" should reflect implementations

[bfgeek]

What is the issue with the HTML Standard?

https://html.spec.whatwg.org/#textarea-wrapping-transformation

From: #6807 (comment)

The "textarea wrapping transformation" doesn't reflect implementation reality (implementations also disagree about how to implement this feature but that's a different conversation).

For this issue the phrase "so that each line has no more than character width characters"

Should be something like "so that each line has no more than the used width of the textarea of content" or something ("characters" a bad term here).

[annevk]

I don't necessarily agree we want to standardize on that. The whole idea is to be able to wrap at a certain amount of characters so you get monospace-compatible wrapping. Making it depend on the width and font in use just makes the whole thing unreliable. Granted, it already is, but then if we don't want to do what it was designed for I'd rather just remove it.

[zcorpan]

Is wrap=hard needed for web compat or would it be possible to remove it?

[hsivonen]

something ("characters" a bad term here).

Gecko uses East Asian Width and General Category as follows (rules higher up on this list win in case of conflicting rules):

U+0000 has width 0.
Other characters below U+0020 have width 1.
C1 controls have width 1.
General categories non-spacing mark and enclosing mark have width 0.
Characters whose East Asian Width is full-width or wide have width 2.
Other characters have width 1.