Free STIX Fonts to be Released in September

tbspit writes "The STIX fonts project has announced that version 1.0 of the STIX fonts should be released in September 2005. The comprehensive font set is to include mathematical symbols and alphabets, and is intended to serve the scientific and engineering community for electronic and print publication. The STIX fonts should be available as fully hinted Type 1 and True Type fonts. The STIX project will also create a TeX implementation. Progress towards release can be monitored here."

Comments

What do they look like?

[Anonymous Coward]

The website talks about how they've been working on the fonts for ten years, but what if they are all butt-ugly? I looked at the website, and there doesn't seem to be even a hint of what they look like. What gives?

Re:What do they look like?

[Cmdr TECO]

Like Times. So yes, butt-ugly, but it could be worse; they could have used Computer Modern.

Re:What do they look like?

[ByteSlicer]

I think they might prove usable. Judge for yourself.

Re:What do they look like?

[Anonymous Coward]

The CM font set is actually well designed.

Re:What do they look like?

[Cmdr TECO]

Yes, it is very well designed -- technically excellent; optical scaling, too. Unfortunately, it is a "modern" (18th/19th century; aka "Didone") typeface, in imitation of the one used in the first edition of The Art of Computer Programming, and suffers from all the faults of that low point of typographic design: extreme contrast, exaggerated round terminals and spindly affected tails, stiff vertical axis, and tiny apertures.

Re:What do they look like?

[Anonymous Coward]

Plus all of the curves (o, c, etc.) are *different* from one another in the italic. Computer Modern Italic looks like it was designed by a four-year-old with a pencil.

Re:What do they look like?

[Cmdr TECO]

Computer Modern's italic(s) is (are) "interesting". The roman is quite close to the typeface used in the TAOCP first edition (I forgot to try to find out what it is exactly), but the italic isn't. The first edition uses conventional italics, for emphasis and in mathematics. In the TeX version (3rd edition of Vol. 1, 2nd for the 2 and 3), Knuth uses a slanted roman instead of italic for emphasis, and the italic used for math is unconventional: although the shapes are based on the originals (i.e. in the "modern" style), all the strokes are very thin, with little contrast -- like a light sans-serif.

(The Computer Modern font files I have on hand are distinctly different from this, so I'm not sure where to put the credit or blame; I don't have the TeX/Metafont books here. I assume some of the egregious flaws in the font in TAOCP 2e, like the "q", were fixed later.)

License?

[general_re]

STIX Fonts Vendor and Publisher License

This license is still in development. It should be completed by June 2005.
...
User License

This license is still in development. It should be completed by June 2005.

Yeah, great, but the devil is always in the details on this sort of thing. If the goal is that "STIX fonts will be made available, under royalty-free license, to anyone, including publishers, software developers, scientists, students, and the general public," let's just put them in the public domain and skip licensing altogether.

But does it support unicode ?.

[Gopal.V]

The biggest pain I have is getting a single font (yeah, I don't mind a 20 MB font that works) which will work uniformly well with unicode text in different languages.

Why the hell don't these people build a single one that really, truly works ?. Until then I'll be using ArialUni.ttf and suffering badly. (texmf is not bad, but the world just doesn't have enough Hellingman).

Re:But does it support unicode ?.

[reidbold]

from the site:

The STIX Fonts have been designed to work with all web browsers, word processors, and other scholarly communications software, as well as all general purpose software.

The Fonts are based on the Unicode(TM) standard for character representation. By expressing all characters with their Unicode value, programs that you use will select the correct glyph for representation.

A character is a unique letter or symbol that is defined by its Unicode value.

Not all Unicode values are included in the STIX Fonts, but there is extensive coverage of Latin alphabets, Greek, and Cyrillic.

So based on the sound of it, this will work in different languages using unicode.

Re:But does it support unicode ?.

[Anonymous Coward]

The biggest pain I have is getting a single font (yeah, I don't mind a 20 MB font that works) which will work uniformly well with unicode text in different languages.

It's going to carry on being a pain, because it's impossible.

Seriously.

There is no way whatsoever that a single font can support English and Japanese text equally well, for example, because Japanese requires that the dots in U+2026 HORIZONTAL ELLIPSIS be placed halfway up the character, while English text requires that they be placed on the baseline. Similarly, the majority of the characters in the CJK Unified Ideographs block need to have different designs depending on whether they are being used in Japanese or Chinese, and in the case of Chinese, they need to be different depending on whether the audience is in Taiwan or the PRC.

The above is not actually true. A combination of a sufficiently intelligent program (with support for langauge tagging and something like OpenType's alternative glyph tables) and a font that implemented all the language variants for all the Unicode characters would technically be capable of rendering any combination of languages adequately.

Why the hell don't these people build a single one that really, truly works?

Because there would be no point in doing so. Without language tagging you couldn't use it, and once you have language tagging it's far simpler just to select an appropriate font for the language.

Re:But does it support unicode ?.

[ulrikp]

The biggest pain I have is getting a single font (yeah, I don't mind a 20 MB font that works) which will work uniformly well with unicode text in different languages.

You mean, like Gentium? Or Doulos SIL?

The Unicode Font Guide for Linux should also give you some pointers.

Ulrik P.

How long does it usually take?

[hritcu]

After nearly 10 years of development, the STIX Fonts project is almost complete.

The community is in great need of such fonts. This open source online equation editor is just an example. We had to recommend the use of a shareware pan-unicode font (Code2000) because the only alternative is the proprietary Arial Unicode MS.

Nevertheless, the time it took them to make STIX almost ready looks hilarious to me. Does anybody know how long does it usually take to design such a font?

Re:How long does it usually take?

[pthisis]

[quote]Nevertheless, the time it took them to make STIX almost ready looks hilarious to me. Does anybody know how long does it usually take to design such a font?[/quote]

Computer Modern was designed over about 12 years.

Of course, Knuth was working on other things (notably Metafont and TeX) in that period as well.

Re:How long does it usually take?

[bcrowell]

Does anybody know how long does it usually take to design such a font?
How long does it take to do an oil painting? Well, it depends on the oil painting. Font design is a serious craft, and this is a big font.

Re:How long does it usually take?

[hritcu]

Unlike an oil painting font design should be to some point paralelizable.

MicroPress has been working on this for the last 5 years and they don't seem like a one-man company to me. After looking at their horribly designed site I am starting to doubt even that. Maybe I'm wrong, but I would have some reticence at hiring them.

Re:How long does it usually take?

[Anonymous Coward]

Most of the work for TeX and MF were done by his graduate students (most notably Andrew Levy). Knuth's involvement was minimal, and generally to the detriment (ie, the TeX programming language).

Re:How long does it usually take?

[Anonymous Coward]

Preview, dude, Preview. This site doesn't use BBS codes, only a limited set of HTML.

Re:How long does it usually take?

[Anonymous Coward]

Most of the work for TeX and MF were done by his graduate students (most notably Andrew Levy).

Really. That explains why a Google search for {TeX "Andrew Levy"} produces a grand total of 34 hits, none of which have anything to do with the typesetting system, then.

Or maybe you're talking through your ass and getting moderated up because Slashdot has a small but vocal anti-Knuth population.

Re:How long does it usually take?

[Phillip2]

Actually, it took them about 5 months to make them nearly ready. The "Status" page went to nearly done that quickly and then has sat there for 5 years.

Very odd.

Phil

Re:How long does it usually take?

[Anonymous Coward]

Actually, Knuth did all the programming stuff on TeX himself. There is this story around that he wrote a specification for two of his students once and "all they had to do" was the implementation. After some weeks of travel or so he returned and saw that they only had a very small prototype ready then. He realized that his specification was not precise enough and most of his ideas were in his mind but not in this spec. He started to write the code himself then.

I hope...

[enrico_suave]

I sure hope it comes with the "Mr. Roboto" font and the "come sail away" font.

Use MathML fonts

[jvj24601]

http://www.mozilla.org/projects/mathml/fonts/

They are freely downloadable (free as in beer), and they have the backing of being used and tested by the Mozilla foundation.

Re:Use MathML fonts

[hritcu]

While awaiting the comprehensive set of fonts being made by the STIX project to cover all the symbols in MathML, use the font installers (on the right) to install the fonts on your system if you do not have them already. MIT has developed convenient font installers for Windows and the Mac, following licensing negotiations through this project and mozilla.org staff (especially considering the open-source nature of Mozilla). The respective font owners have made provision for the fonts to be packaged into these installers, with the aim of helping to boost the adoption of MathML into the mainstream.

MathML-enabled Mozilla uses the MIT fonts, but it first maps them to the right entities. This happens in the code because the fonts, although free, are not to be touched or redistributed. Without the right mapping the fonts are useless, and for anything other than standalone applications you cannot perform such a mapping. So I think that you might be forgetting that the main focus of MathML is the Web not standalone applications. The CSS "font-*" attributes don't allow characters to be mapped to different fonts so I doubt that the MIT fonts are of any real use on the Web (unless you are targeting only the users of MathML-enabled Mozilla).

Bold

[LordMyren]

I'm having some major issues with Bold fonts on my laptop. They simply dont look bold. I love Sans, but it refuses to look at all different in bold.

My systems' DPI is 133, I have to give gnome a font size of 4-5 at that DPI to get the small size I want. I just lie and tell it my DPI is 66, and set the fonts to 9 and 10. But the fonts still dont "look" bold. I honestly cannot distinguish which of my Liferea feeds have been updated and which havent (bold/not bold).

I tried a whole army of fonts, still no dice. Its my understanding that you can hint bold fonts. My recommendation to STIX? Provide some extra bold hintings as well as just bold, and make sure they make very small fonts actually look bold.

The main app I try to configure with is the gnome control center's font selection (well done guys, ::pat on back::). I dont know any better.

Re:Bold

[Anonymous Coward]

You could use a utility called gtf, which comes with xorg/xfree (I guess).
Now you should have a optimal DPI according to your monitor specs and chosen resolution.

$ gtf 1280 1024 85

# 1280x1024 @ 85.00 Hz (GTF) hsync: 91.38 kHz; pclk: 159.36 MHz
Modeline "1280x1024_85.00" 159.36 1280 1376 1512 1744 1024 1025 1028 1075 -HSync +Vsync

Re:Bold

[Anonymous Coward]

It sounds like your fonts are messed up. With Bistream Vera Sans on my system (What most systems point the Sans alias to), I can tell if the font is bold even when the letters themselves are illegible.