On Mon, Aug 24, 2020 at 10:01 AM Jonas Smedegaard <
jonas@jones.dk> wrote:
Quoting Fabian Greffrath (2020-08-24 10:21:09)
> Am 2020-08-24 10:08, schrieb Jonas Smedegaard:
> > ...and it would be very sad to not provide them to our users!
>
> Because why? What advantage do they provide compared to separate font
> files for each flavor?
Because variability across one or more axis provide far more resulting
fonts than a set of specific compositions.
VF files are also more compact; certainly when there are > 2 styles, and the savings can be quite drastic where, for example, CJK fonts are concerned.
Also, Pango and GTK support them in addition to the packages mentioned earlier, and there was some work showcased that exploited the variation feature as a UI/UX element, such as having text labels in GTK widgets grow bolder upon mouseover, so eventually I would expect them to become the norm.
Hypothetically, there might be software stacks that don't work with them, in which case dealing with those would be a target for specific workarounds.... I am not certain, for example, whether or not all printing backends can generate "expected" PDF or PS output from variable fonts with customized variation settings. Yet. But my impression was that that will eventually be fully fixed, for all actively-developed libraries. I dunno whether the same can be said or Tcl/Tk or whatnot....
I agree we should not package exact same font multiple times (i.e. let's
throw away woff fonts, as soon as we have established a mechanims to
auto-generate those for web apps needing them).
This is a good question; I seem to recall that some web-application mega-service packages bundle in WOFF fonts. Perhaps that is just out of inertia/lack-of-interest or accident, however, rather than a real requirement?
If we want fewer font packages, then let's throw away conventional
fonts, not variable fonts! But let's wait a bit before we throw away
conventional fonts.
It is possible to use update-alternatives to offer users a choice in this sort of situation? I am actually curious what the community temperature is about that as an approach — if it causes more confusion that it does solutions, or if it's regarded as an extreme-measures type of packaging tactic.
I 100% support making variable fonts the default; I only ask about fallback approaches.