From: Peter Kirk (peterkirk@qaya.org)
Date: Mon Aug 09 2004 - 16:58:08 CDT
On 09/08/2004 22:10, Mike Ayers wrote:
> ...
>
> > expect any kind of break or control character to have this
> > effect by default. The correct way to do this is probably to
> > insert SPACE or NBSP, or the proposed INVISIBLE CHARACTER.
>
> ...except that I don't want a space in there, nor can I use a
> proposed character.
>
Well, the only specified way which works for all characters is to use
SPACE or NBSP. Several people including me have seen a problem with
this, and presumably this is why INVISIBLE LETTER has been proposed,
http://std.dkuug.dk/jtc1/sc2/wg2/docs/n2822.pdf.
> > And then of course there is the spacing clone of the macron, U+00AF.
>
> Possible. Do such clones exist for all combiners?
>
No.
> > I wouldn't consider this a violation of layering. If layer A
> > has to split a string at a particular place because of its
> > own functions, that does not imply a break at layer B.
>
> ...if layer A is on top of layer B, then a break in layer A
> is, by definition, a break in layer B. Could I be using the wrong
> term again?
>
Well, maybe, but this is certainly not true "by definition" of all kinds
of layering. In this case layer B simply doesn't recognise breaks in
layer A. For example, layer A might include markup within <...>, but as
far as layer B is concerned < and > are ordinary characters which are
part of a character stream and not recognised as markup. But that is not
the relationship we have here, which is not a simple layered structure.
> > And it
> > is certainly a legitimate user requirement for a combining
> > mark to be in a different colour from its base character.
>
> Has this been accepted by Unicode as a requirement? If so,
> I'll just run screaming away from the issue.
>
It is probably considered outside the scope of Unicode. Nevertheless, it
is something which people commonly want to do.
> ... Color change is the only markup that I can think of for which it
> would be possible to preserve kerning across markup based change, ...
>
I deliberately mentioned underlining as another example. And then there
are all sorts of markup which do not affect rendering and so need not
interrup kerning. Just one example: language and script tagging. And
believe me, I have seen words which change script in the middle!
> ... and I assert that "people who change font colors midword" and
> "people who care about kerning" are two sets which do not intersect.
>
>
Don't count on it. Some software makes good use of font colour changes,
which could be mid-word. That does not imply that kerning should be broken.
I'm not suggesting that all rendering engines must have such
capabilities, just that there might be real requirements for such things
for certain purposes.
-- Peter Kirk peter@qaya.org (personal) peterkirk@qaya.org (work) http://www.qaya.org/
This archive was generated by hypermail 2.1.5 : Mon Aug 09 2004 - 16:59:33 CDT