Re: Font dependencies

[Aleksej Saushev <asau%inbox.ru@localhost>]

David Holland <dholland-pkgusers%netbsd.org@localhost> writes:

> On Wed, Jun 20, 2012 at 05:23:37PM -0500, Tim Larson wrote:
>  > > There should also be at least one font meta-package that installs decent
>  > > basic fonts for all alphabets.
>  > 
>  > A meta-package per "alphabet", pointing at a Free font that
>  > provides full (or the fullest available) coverage for that
>  > alphabet, might not be a bad idea.  A meta-package for "all
>  > alphabets" would point at these, in turn.  Which font to use is
>  > open to discussion, of course.
>
> I was thinking that one meta-package would be enough; people who
> wanted to know what the recommended default Thai font is could look at
> the meta-package contents and then install the font they wanted
> directly.

You're missing the point completely.

You _cannot_ fix it that way.

Firefox required exactly Deja Vu while Emacs worked just fine with default 
fonts.
You can install only two fonts, fonts/font-misc-misc and fonts/font-cursor-misc,
and these are sufficient to work with Cyrillic texts. In Emacs (and Gnus).
But not in Firefox and Thunderbird, as it turns out.

All the text below is theory with no experimental evidence.
(It is unrelated to original problem either.)

>  > I would put forth that a good metric might be, use the font that
>  > fully covers the most alphabets, so that at the end of the day the
>  > least amount of time and space is used for the installation.  If we
>  > are concerned with 50 alphabets, and there's a font that fully
>  > covers 40 of them, that's a good candidate to choose.
>
> Well... no, I don't think we should default to a low-quality font just
> because it has wide coverage. For any particular alphabet we should
> pick a font that won't embarrass the project.
>
>  > These meta-packages might have a list of all fonts (in pkgsrc) that
>  > do provide full coverage of the alphabet in question, so that if
>  > one that is not the "top choice" IS installed, the meta-package is
>  > satisfied and does not try to install anything.  Obviously this
>  > would need care and maintenance from typeface geeks.  Possibly some
>  > tool could be written to auto-update this font coverage info
>  > whenever a font package is added/changed/removed.
>
> We do have alternatives logic built into the depends system, so in
> theory one could have a package called e.g. any-hebrew-font that
> depended on one expression containing all the available such fonts as
> alternatives. But this expression would be horrific where lots of
> fonts are available, such as for latin1.
>
> I was thinking it would want to be some mechanism like requires/
> provides, as in "this package provides (a) font(s) for: greek" and
> "this package requires a font for: greek".
>
> But then, it's also not really alphabets as such either, it's glyph
> sets, and even if we assume that this can be adequately expressed as
> subranges of the Unicode glyph space (which is not necessarily the
> case) there are a lot of complications that might appear. What about,
> e.g. cyrillic fonts that don't include the letters used only in
> Ukrainian? I'm sure some of these exist as relics of the Soviet era.
>
> There's nothing easy about proper i18n unfortunately...


-- 
HE CE3OH...