Re: [1.2T-LC] Glyphs and <text> offsets (ISSUE-2169)

Doug Schepers wrote:
> Hi, fantasai-
> 
> fantasai wrote (on 10/30/08 4:58 PM):
>> http://dev.w3.org/SVG/profiles/1.2T/publish/text.html#CharactersAndGlyphs
>> http://dev.w3.org/SVG/profiles/1.2T/publish/text.html#CharactersAndGlyphs
>>
>> Given that whether 'fi' is a single glyph or two glyphs depends on the
>> fonts available, and given that the mapping of the <text> element's
>> coordinates to letters depends on the number of glyphs, does this mean
>> that assigning coordinates to letters in a word like
>>   final
>> depends on what fonts are available?
>>
>> If so, that seems like a problem.
> 
> This is functionality that has been in SVG since the beginning, and it
> doesn't seem to have caused problems.  Indeed, authors do need to be
> careful about specifying fonts when using glyph positioning, but with
> SVG Fonts (and the downloadable Web fonts that will hopefully be
> available soon through other W3C initiatives), authors can expect to
> have a reasonable amount of control over this.
> 
> Still, this is something we could revisit in SVG 2.0 Core, if necessary.
>  Perhaps we could indicate that glyph positioning negates the use of
> ligatures; there is precedent in languages like TeX, which provides a
> way to suppress undesired ligatures.  Another option would be a specific
> property that controls whether ligatures are used; this might also be
> desirable for letter-spacing.
> 
> While the SVG WG doesn't feel that there is a pressing need to address
> this in SVG 1.2 Tiny, we are happy to work with the CSS and XSL WGs in
> the future to solve this quirk.
> 
> Please let us know if this satisfies your comment.

I will consider my comment satisfied for SVG Tiny 1.2. I'll note for
the future, that you may want to map coordinates to grapheme clusters
rather than to glyphs and let the UA skip coordinates that would map
into the middle of a multi-grapheme-cluster glyph. That way, since
grapheme clusters are defined by codepoints rather than glyphs, you
will get a consistent coordinate-character mapping across platforms
and implementations.

(I'll note this isn't an issue for CSS because we don't have this
capability. The problem was just something I noticed in passing while
reading Richard's i18n comments.)

~fantasai

Received on Friday, 31 October 2008 17:37:17 UTC