[Pkg-fonts-devel] font review updated

Paul Wise pabs at debian.org
Mon Dec 10 23:07:36 UTC 2007

On Dec 11, 2007 12:17 AM, Nicolas Spalinger <nicolas_spalinger at sil.org> wrote:

> Did you also set up a regular cron setup? Like a weekly regeneration?

Nope. I didn't know at the time that fttools is installed, but I see
now that it is. I'll set this up today and make it run just after the
Contents file regeneration.

Later we'll need fontforge too for converting more font types and for
unicode coverage info.

> > I'm also thinking of integrating the work that Gürkan Sengün did on
> > [1] and adding Unicode coverage information using fontforge and
> > DejaVu's unicover.pl
> >
> > 1. http://krum.ethz.ch/fonts/
> What features are you thinking about?

From krum? Mainly better previews, support for more varieties of fonts.

Some other ideas:

Better separation of packages/fonts into main/contrib/non-free

Detecting duplication based on font family overviews

Unicode block and script coverage

Blacklisting files from Contents that are known to be symlinks

Better layout & different sized previews

Put all the overviews on index.html

Cleaning up the ftinfo output

Some stuff about the private use areas of unicode.

Maybe some RPM support for the fedora folks.

> I don't see how having a huge timestamp-ed snapshot of all the fonts in
> the archive in huge tarball can be useful. There are no corresponding
> descriptions and documentation. Rather misleading about the specifics of
> each font family and potential problems. IMHO reminiscent of "freeware"
> font sites where folks offer thousands of "free fonts" in one go.

I agree with this and won't be adding that feature.

I think the idea was to make all these free software fonts available
to Windows users at Gürkan's workplace.

> The great work done by Myriam had the goal of helping us reduce the
> duplication in the packages in the archive, help us review where the
> metadata was wrong so we could communicate with upstream, unify the
> licensing and spot non-free fonts who have slipped in.

Yep. MD5-sums are good, but there are several Vera/DejaVu/FreeFont
versions in the archive from what I can see after looking over
fonts.txt. I want to add some way to detect these multiple versions,
perhaps some overviews of font families and variants.

> But yes I agree that we need to work on adding Unicode coverage (5.0 and
> 5.1 later on) for the fonts, the new Dejavu build script should help
> (there's also a ttfcoverage script using Font::TTF in my folder that we
> can hook up to the review script).

I'd prefer to use DejaVu's unicover.pl or something similar since we
can convert other fonts to .sfd and then use unicover.pl on all of
them. This way we can use the latest Unicode block/script info from
the Unicode website rather than relying on the block/script info in
any package. I also wonder how the text files in unicode-data can be
distributed in main since they seem non-free to me.

I also wonder about using fonttools for this since it knows about a
few types of fonts.

> Using the latest ftview (from the package freetype2-demos) will allow
> for better internationalized output and to see all the glyphs contained
> in the font instead of a latin-centric pangram.

The sid version of that doesn't seem to allow writing to files, unless
there is an undocumented option or I'm miss-reading the --help output.

> Ideally we should integrate our review script with the great work Davide
> has done for d-i font rendering reviews:
> http://d-i.alioth.debian.org/gtk-frontend/screenshots/

Indeed, I'm not sure what such integration would entail though.



More information about the Pkg-fonts-devel mailing list