Re: [css3-lists] Publish a new WD?

On Mon, Nov 21, 2011 at 4:39 AM, Håkon Wium Lie <howcome@opera.com> wrote:
> Based on my comments I would suggest listing these issues in the draft:
>
>  - Issue: Is it possible to find a syntax for several list markers to be
>   written in one string? One possible solution is:
>
>    @counter-style lower-norwegian {
>      type: alphabetic;
>      glyphs: 'abcdefghijklmnopqrstuvwxyzæøå';
>    }

I've now added this as an issue.


>  - Issue: Do we need predefined lists beyond what CSS2 defines? For example,
>   do we need 'simple-upper-roman', 'fullwidth-decimal', 'octal',
>   'upper-hexadecimal'? Do the people who need octal numbering (there
>   may be some) really trust CSS to get numbering correct?

This is already covered by an issue.  (I'll note, though, that 'octal'
isn't a very good style to question the correctness of, given that
it's completely trivial.  Better would be some of the non-English
alphabetic styles, where I'm counting on information from other people
and my own transcription ability to get this right.)


>  - Issue: should we require real-world examples of all list style
>   types described in this specification?

Most of the styles are stated to have real-world usage in the various
emails or documents that I and Hixie extracted the original styles
from.  I can produce those original emails, but it would be a *lot* of
work to go through and re-justify all of them.  I'm not particularly
interested in doing so.

We could trim the fat a little, but it would remove only a handful or
two of styles.


>  - Issue: If we decide that we need more predefined list types, what
>   criteria should be used and how should that criteria be expressed
>   in the specification? Does presence in Unicode warrant placement?
>   Do we need lists for Norwegian, Swedish, Danish and other languages
>   written in the Latin scripts. If not, why not?

The (unstated) criteria that I used was if it's a living language and
the usage was stated to be reasonably popular.  Do you want something
in the spec about this?


>  - Issue: Should 'footnotes' be 'footnote' instead? (like CSS does in
>   "italic" and other places)

Perhaps.  Do you have a strong opinion either way?  I don't.  If you
make a decision here, we don't need an issue over it.


>  - Issue: should we replace the numbering systems described in chapter
>   11 with spelled-out lists that can be expressed without defining
>   algorithms? Before deciding, spelled-out lists up to, say 100, should be
>   added for comparison purposes.

Up to 100 isn't really acceptable.  While *most* lists are under 100,
as they're hand-coded, a significant fraction of the remaining
use-cases can get very large, easily going into the thousands.

That said, I don't speak Chinese or any of the african languages that
would use ethiopian-numeric, so I have no idea if the chapter 11
styles are expected to be used on the types of lists that'll get
numbered into the thousands.

I've filed an issue on this.


>  - Issue: how far do we need to count? Unicode defines list makers up
>   to 20 (e.g. for parenthisized numbers, full stop numbers, and
>   circled numbers) Do we need to go further? If so, how far?

If it's a unicode-based style, we obviously can't go further.  I've
defined that UAs may optionally synthesize glyphs past the unicode
limits, but I don't think we need anything stronger.


>  - Issue: how do we know the algorithms are correct?

Review, hopefully!


>  - Issue: should the example in 8.1.4 be marked up as an example?

That paragraph has no conformance keywords, and has a parallel
structure with the rest of the definitions where that paragraph is
just an introductory explanation of the type.  I added it there only
because you didn't like the "doubling" terminology.

That type already has a similar example actually marked up in an example block.


> And a new one, not discussed in my previous comments:
>
>  - Issue: could W3C host a style sheet with the "predefined" styles in
>   it? It's easier to correct errors in this style sheet than it is to
>   change/update deployed browsers.

Given the pain caused by software actually following doctype urls that
pointed to the w3c, I doubt this would fly (and software is *supposed*
to follow <link rel=stylesheet> urls!).  ^_^

If any errors linger past CR, users will find them and they should
hopefully filter back to us reasonably quickly, so we can fix them
without much compat issues.  As well, users can fix them for
themselves temporarily.

~TJ

Received on Monday, 21 November 2011 23:26:28 UTC