Re: ligature formation across text chunks

s/non-discontiguous/discontiguous/

On Fri, May 13, 2011 at 8:33 AM, Glenn Adams <glenn@skynav.com> wrote:

> Alex,
>
> In my message, by glyph index I did not mean the glyph identifier or code
> used in the font, but rather the index in the sequence of glyphs in the
> example I referenced. In any case, it does seem that SVG can support
> referring to glyph identifiers via the glyphRef attribute or element.
>
> Part of the point of my email was also to remind the group that one can
> have a discontiguous set of characters in the original input character
> sequence that maps to a single ligature glyph. One can also have one
> character mapping to multiple discontiguous glyphs. In the general case, one
> has an M to N mapping from characters to glyphs, where both M and N may be
> non-discontiguous.
>
> Regards,
> Glenn
>
>
> On Thu, May 12, 2011 at 11:50 PM, Alex Danilo <alex@abbra.com> wrote:
>
>> Hi Guys,
>>
>>        I raised this a while back. It's tracked as an issue:
>>
>> http://www.w3.org/Graphics/SVG/WG/track/issues/2332
>>
>>        I think the behaviour described by Vincent is more likely to be
>> useful
>> to authoring tools.
>>
>>        For example, HP's PCL-XL explicitly positions each glyph for
>> printing. Microsoft's XPS does the same. The usefulness of the X,Y
>> positions
>> is more than likely for accurate glyph placement when seriallizing output
>> from
>> some tool (like SVGMaker for example). In such cases, the ligatures are
>> more
>> than likely wanted.
>>
>>        As a case of how to do things properly(;-) XPS defines a set of
>> glyph
>> indices with matching X, Y positional information - then provides the raw
>> Unicode
>> string as another attribute so that character copy/paste etc. can work
>> using the
>> content in the file.
>>
>>        SVG does not have the concept of glyph indices, so we're stuck with
>> the
>> character string only. So, to get precise glyph positioning (including
>> ligatures), it
>> must be done via X,Y or dx, dy.
>>
>>        There are tests in the test suite to check this behaviour IIRC.
>>
>> Alex
>>
>>
>>
>