Re: real vs. synthetic width glyphs



On 7/8/13 6:25 PM, "fantasai" <fantasai.lists@inkedblade.net> wrote:

>On 07/08/2013 03:48 PM, Sylvain Galineau wrote:
>>
>> Florian wrote:
>>>
>>> But when all the glyphs are available, leaving some wiggle room to the
>>> implementation seems counter productive if the only way they can
>>>deviate
>>> from our preferred behavior is by being worse.
>>
>> Right. To use small-caps as an analogy: if the font has the small caps
>>glyphs
>> you need you're supposed to use them. But if they're not present there
>>is no
>> strict definition of what fallback you should use. css3-fonts only says
>>UAs
>> 'should simulate a small-caps font, for example by…' scaling uppercase
>>glyphs.
>> This leaves the door open to UA innovation when the type designer
>>didn't do the
>> job.
>
>This isn't quite the right argument here. The author isn't
>requesting half-width glyphs. The author is requesting
>that these glyphs be combined and made to fit within 1em.

If the author is requesting a small chunk of text to be laid out in a
manner for which the font he has *chosen* includes specifically designed
glyphs I think the small-caps analogy is apt. I do not believe we claim
the UA should make up those glyphs for him using proprietary magic.

>
>Sometimes half-width glyphs is the right way to do that.

We do not choose defaults for what happens sometimes; we pick them based
on what happens most of the time. For tatechuyoko my understanding is that
a short string of digits is by far the most common use-case.

>Sometimes you get better results just with proportional-width
>glyphs, because of differences in the width of the glyphs.
>Half-width glyphs have a mono-space characteristic; narrow
>characters are made to look wider, wide characters squashed
>to be narrow. If you combine a narrow character with a wide
>one, sometimes that fits within 1em without the squashing
>and stretching, and that result looks better than flipping
>into a monospaced glyph set.

You are describing an hypothetical edge-case. If the main use-case  is
natively supported by the font then this is what UAs should use by default.

For more complex typesetting, additional tweaking may be needed, and that
is perfectly fine. It seems a bad trade-off, however, to allow UAs to
produce inconsistent and/or poor quality results for the main use-case.

>
>~fantasai
>

Received on Tuesday, 9 July 2013 06:43:25 UTC