Re: [css3-fonts] Variants (Ligatures, Capitalization, Digits)

On Thu, Mar 10, 2011 at 2:27 AM, Christoph Päper <
christoph.paeper@crissov.de> wrote:

> Thomas Phinney:
> > On Wed, Mar 9, 2011 at 9:02 AM, Christoph Päper <
> christoph.paeper@crissov.de
> >
> >> * Section 6.7 Numerical formatting
> >>
> >> And I still believe the value names are not well chosen, and neither are
> they in Open Type:
> >>
> >> “lining”       (+lnum, -onum)
> >> “casing”       (+onum, -lnum)
> >> “proportional” (+pnum, -tnum)
> >> “monospaced”   (+tnum, -pnum)
> >>
> >> “oldstyle”     (+onum, +pnum)
> >> “tabular”      (+tnum, +lnum)
> >>  …
> >
> > Where are you getting that mapping of feature combinations to labels?
>
> Nowhere but my own preference. I like clear antonyms better than some
> arbitrary, traditional names that are not even used consistently in the
> wild.
>
> > I'm sure I'm missing something here.
>
> Nope.
>
> > The definition of "oldstyle" seems odd in requiring proportional….
>
> Is it? Judging by the OT feature names, it is, but if you forget that for a
> moment, does “old style” actually give any hint about digit width, height,
> both or neither? No, it’s completely arbitrary, whereas “tabular” at least
> describes the most common use case, while “lining” and “proportional” are
> derived from inherent properties. I’d like all keywords to be of the latter
> type, because it lessens the burden to learn some jargon. I am not against
> using the other terms as aliases for (combinations of) basic features, hence
> their inclusion above; preferably there were also names for (+onum, +tnum)
> and (+lnum, +pnum).
>

Not so much as pre-existing terms in English AFAIK, no.


> Ideally, but perhaps not realistically, there were common terms for the
> ‘font-variant’ shorthand property that set the sub-properties to sound
> values, without exposing the details there directly. That means, e.g.
>
>  font-variant: titling;
>
> would result in something like
>
>  font-variant-caps: titling-caps;
>  font-variant-numeric: lining-nums proportional-nums;/*=titling-nums*/
>  font-variant-ligatures: common-ligs no-historical-ligs;/*=titling-ligs*/
>  font-variant-alternates: normal;
>  font-variant-east-asian: full-width;
>
> or
>
>  font-variant: historic;
>
> would set
>
>  font-variant-ligatures:  historical-ligs;
>  font-variant-caps:       normal;
>  font-variant-numeric:    oldstyle-nums;/*=casing-nums proportional-nums*/
>  font-variant-alternates: historical-forms;
>  font-variant-east-asian: traditional full-width;
>
> etc.
>

I don't object in *theory* to having shorthand ways of setting more than one
characteristic, whether it's two or a bunch. But I think I object in
practice because I expect it would get confusing.

If I were thinking like Liam, I might be concerned that this is subject to
the same issue as having high-level and low-level ways of setting an
underlying property. This means that there could be *three* ways of doing
certain things, which seems to be even more potentially confusing.
Especially if the names give an incomplete picture of the effects of a
setting. If I have set tabularity on I might be surprised when setting the
oldstyle figures "the wrong way" also switches the numbers back to
proportional....

Regards,

T

-- 
“Puritanism: The haunting fear that someone,
 somewhere, may be happy.”
 —H.L. Mencken