[css-text] Hanging/collapsing of fixed-width spaces not specified

[briansmith]

https://www.w3.org/TR/css-text-3/#white-space-rules says "White space processing in CSS affects only the document white space characters: spaces (U+0020), tabs (U+0009), and segment breaks."

However, it is often/usually desirable for the collapsing/hanging behavior to apply to fixed-width spaces when the fixed-width space would cause a line break. Consider:

<p>1&#x205f;+&#x205f;1&#x2005;=&#x2005;2</p>

Which renders as:

1 + 1 = 2

The spaces on either side of the + are U+205F Medium Mathematical Space and the spaces on either side of the = are U+2005. (This is the spacing recommended in http://www.unicode.org/notes/tn28 for these operators.)

Lines wrap at fixed-width spaces already (at least in the browsers I tested). However, in the browsers I tested, these fixed-width spaces do not currently collapse or hang at the end of a line like U+0020. Either collapsing/hanging of these spaces like U+0020 should be the default behavior, or there should be some way to opt into them being treated like U+0020 for the purposes of collapsing/hanging.

[frivoal]

Doing something like this could mean changing either or both of these sentences in 4.1.2 to cover more types of spaces, (maybe U+2000 to U+200A and U+205F)

• A sequence at the end of a line (ignoring any intervening inline box boundaries) of collapsible spaces (U+0020) and/or ideographic spaces (U+3000) whose white-space value collapses spaces is removed.

• If there remains any sequence of white space and/or ideographic spaces (U+3000) at the end of a line:
  [... rules about pre-wrap and break-spaces..]

Conceptually, that makes sense to me.

[litherum]

There's like a million billion different flavors of spaces in Unicode. What was the reasoning for not including them since their inception?

[fantasai]

@litherum Legacy, I imagine. We can gather them up and call them by category Zs, fwiw. http://www.fileformat.info/info/unicode/category/Zs/list.htm Hanging Ogham Space Mark is a bit odd, though, since it's usually visible.

[css-meeting-bot]

The CSS Working Group just discussed Hanging/collapsing fixed-width spaces, and agreed to the following:

• RESOLVED: fixed width spaces U+2000 and up are treated like U+0020 and U+3000 -- hang / disappear at end of line

The full IRC log of that discussion

<fantasai> Topic: Hanging/collapsing fixed-width spaces
<fantasai> github: https://github.com//issues/3879
<fantasai> florian: Spaces ...
<fantasai> florian: Currently only U+0020 and U+3000 are able to hang
<fantasai> florian: But other fixed-width spaces don't
<fantasai> florian: don't know if there was a specific reason not to include them, or some kind of oversight
<fantasai> fantasai: I think we knew that nbsp couldn't hang
<fantasai> fantasai: so we only had U+0020 hang
<fantasai> fantasai: later on added ideographic space in response to feedback from Japan
<fantasai> fantasai: Didn't re-evaluate other fixed-width space
<fantasai> florian: Compat risk?
<fantasai> fantasai: Probably low, most people don't know they exist
<fantasai> fantasai: I'd be willing to change the spec and see how it goes
<fantasai> fantasai: We'll need to except nbsp
<fantasai> fantasai: also ogham space mark?
<fantasai> florian: There's about whether spaces can hang at the end of the line in pre-wrap
<fantasai> florian: separate question of whether they disappear when wrapping 'white-space: normal'
<fantasai> florian: Not sure how to deal with ogham
<fantasai> fantasai: will have to as i18n
<fantasai> RESOLVED: fixed width spaces U+2000 and up are treated like U+0020 and U+3000 -- hang / disappear at end of line
<fantasai> http://www.fileformat.info/info/unicode/category/Zs/list.htm

[briansmith]

fantasai: I think we knew that nbsp couldn't hang

I'm not disagreeing, but I don't understand that is special about nbsp that makes it necessary for it to be excepted.

[fantasai]

@briansmith It has a lot of legacy usage in holding space for various layouts, so changing its behavior is very risky from a Web-compat perspective.