A survey of free font licenses (NewsForge) [LWN.net]

Two parts to this

Posted Mar 1, 2006 7:01 UTC (Wed) by jd (guest, #26381) [Link]

First, there are many TrueType, but very few OpenType, fonts. There are many Type 1, but very few Type 3, fonts. There is a definite lack of fonts that make use of the more modern kernings and capabilities. A good overhaul of what already exists would therefore be no bad thing.

Second, there are lots of fonts for "common" alphabets. The market is glutted with those. However, this is not where the Really Big Interest in Open Source seems to lie. If you look at the countries adopting Open Source on a grand scale, they are not the major ones (although Britain and Americs are doing OK in that regard). The countries likely to adopt Linux as the OS of choice in education in the future will more likely be found in Africa than America.

Further, who are the main adopters of Linux in the Western world? Secretaries or scientists? My money is on the latter.

On those grounds, I would argue that SIL should emphasise the rarer languages and languages from very poor nations, as those will have maximal educational impact. For scientists, they should also put effort into languages and useful symbols that are extremely difficult or impossible to obtain at low cost.

More of the same won't add any value to what's already there, even if it is under a Free/Open License. It's hard to debug an umlaut and if you really want to, you just run a converter over a metafont. The highest priority would seem to need to be placed on those fonts this is impossible for.

Two parts to this

Posted Mar 1, 2006 7:22 UTC (Wed) by Ross (subscriber, #4065) [Link]

Your mentioning of different font types, kernings, and scientists reminded me of TeX. That format is supposed to support some very advanced capabilities and there are a large number of fonts in it (though they are in an unusual character set which is neither a subset nor superset of Unicode). Is there a technical or legal reason that Metafont fonts can't be used for display?

Two parts to this

Posted Mar 1, 2006 9:53 UTC (Wed) by rsidd (subscriber, #2582) [Link]

TeX users themselves are increasingly using Type1 fonts rather than
METAFONT fonts. One major reason is certainly that there are more Type1
fonts available. But also, METAFONT was probably too sophisticated for
most people: unlike other vector fonts it's not an outline font, but uses
finite-width "brush" strokes. It's much slower render than Type1 and
TrueType fonts. And there is no standard rendering engine other than the
METAFONT program itself. So TeX makes bitmap fonts at the desired
resolution from the METAFONT descriptions, and uses those in its "DVI"
files. That was great in the early 1980s when there was no font
standardisation, but now even people who use Computer Modern (TeX's
"native" fonts) usually use the BlueSky Type1 versions of those fonts, so
that they can be embedded into the output (PostScript or PDF) files.

There are many cool things about the METAFONT system but they're
outweighed by its impracticality.

As a print-quality typesetting system TeX is still unmatched in technical
typographic quality -- not just in mathematics, where nothing else comes
remotely close, but even in regular text. But, largely because of the
font handling, it's a poor fit for general-purpose text formatting.

Two parts to this

Posted Mar 1, 2006 11:04 UTC (Wed) by hanwen (subscriber, #4329) [Link]

the slowness or clumsiness of METAFONT isn't really a problem for printing. There are tools (eg. mftrace which I wrote, http://www.xs4all.nl/~hanwen/mftrace , fontforge), that generate Type1/TTF fonts from METAFONT code. The quality is more than sufficient for hires printing.

Unfortunately, for display fonts, the outlines need to be of high quality, and also contain good hinting information, which fontforge can't produce yet. Moreover, a lot of TTF/OTF fonts actually contain separate bitmap images for use with low resolutions, which (I suppose) are hand-tuned.

Should be solvable.

Posted Mar 2, 2006 0:11 UTC (Thu) by jd (guest, #26381) [Link]

Metafonts are written as scriptlets. Bits of actual programmable logic. For web work, all it should require is a plugin that treats the metafont code as an embedded script and the web page as the data on which the script should be run.

For GUI interfaces, you'd want to write a highly accelerated rendering engine which the existing font managers can link to. Leave the TeX font rendering software alone for this - it's good for what it's designed for, but this is a different type of application. ALTERNATIVELY, modify one of the TeX font rendering engines (there are two, I believe) to generate bitmaps compatible with the original X bitmap format. You lose all the benefits of scalability like that, but nobody uses scaling much anyway.

(For those who say "huh?" to that last part - if you make a window larger and have a way of stretching the characters so that you get the same amount in the window but at a higher quality, you've made use of scaling. If you can merely select a specific size, there's no difference between scaling and having pre-generated bitmaps.)

Metafonts suffer some from aging - I don't believe the format has been updated much since it was originally designed, but the capabilities of devices and the needs of users have changed dramatically. Again, it would be very useful - in my totally egocentric opinion - if Metafonts could be extended to reproduce all of the features in modern fonts, in addition to looking at the needs of LaTeX 3 and looking at the wishlist of font designers and font users that don't think existing font systems do what they want.

If Metafonts could become a superset, then conversion to and from them would be extremely easy AND Metafonts could be proposed as the universal font sourcecode exchange format - which, whether you use metafonts in actual applications or not - would be an excellent and valuable use of metafonts.

Two parts to this

Posted Mar 1, 2006 11:06 UTC (Wed) by hanwen (subscriber, #4329) [Link]

A Type3 fonts is not a desirable thing. "Type3 font" is a name for a collection of generic postscript routines. They are slow to render and impossible to hint.