OS limitations lead to separate OT-fonts?

Christoph's picture

On P22's blog it is said

Due to some OS limitations, the Pro fonts will be available separately as Latin, Cyrillic and Greek

What could that mean?

guifa's picture

Given the nature of that font, I can't imagine that there's anything OpenType-related difference in operating systems that could cause a problem. Especially since it would really be the font renderer, not the OS, that would have problems with OpenType-ness, right? Unless maybe there's something odd about the font file that causes the default font managers to have some trouble, but that sounds a bit fishy.

I vote for they just want three times the money. It makes sense though, as they say it's their largest font by glyph count, and the average Western European will just be using the Latin script. While he might only use that, he'd still be getting three times more than he'll use, and so if they didn't split the font into three version, they'd rightfully charge two or three times more for their font. So, perhaps it's being nice to those users who only need one of the three.

That's just a guess though. I do call BS it being an OS limitation.

«El futuro es una línea tan fina que apenas nos damos cuenta de pintarla nosotros mismos». (La Luz Oscura, por Javier Guerrero)

Nick Shinn's picture

I'd say it's probably a mental slip, and they meant "application limitations".
For instance, Quark 6.5 or earlier.
That app no longer works on OS X since Mac switched to Intel and there's no "Classic mode".
And in Quark 6.5, although OpenType fonts do work, there is no feature support, which is what the "Pro" designation is all about.
So in a sense it's an OS limitation, in that OS X doesn't support older Quark.

Miguel Sousa's picture

> Given the nature of that font, I can’t imagine that there’s anything OpenType-related difference in operating systems that could cause a problem. [...] I vote for they just want three times the money. [...] That’s just a guess though. I do call BS it being an OS limitation.

> I’d say it’s probably a mental slip, and they meant “application limitations”. So in a sense it’s an OS limitation, in that OS X doesn’t support older Quark.

And I say, you two DON'T KNOW anything about what you're talking about!! (I hate when this happens!!) Has it occurred to you to do some research, including e-mailing the foundry, before jumping into conclusions?

The P22 London Underground font triggers the same OSX bug that was found when Arno Pro was being developed:
http://docs.info.apple.com/article.html?artnum=304821

Although huge, Arno's glyphset is dozens of glyphs smaller than P22 London Underground, which made it possible to find a workaround for the above OSX bug. Unfortunately, the same was not possible to do with P22's font, despite Paul Hunt's efforts.

Miss Tiffany's picture

Before we go and have another ugly thread I think it is important when you have a question like this you email the foundry. Not everyone is looking for more money.

kegler's picture

I vote for they just want three times the money...I do call BS it being an OS limitation.

No, actually it is three times the work to separate them out since all of the fonts are available as sub-families. The end price of the full font system is the same as originally planned. If someone wants just one script, they will actually be paying LESS. Because of stylistic variants increasing the glyph count to over 5000, there are problems with earlier OS behavior. This is not a marketing scheme or gimmick. Unless we demand all of customers have the latest OS, there are some issues. It has nothing to do with application support of OT features. The fairest workaround for the widest customer usage ended up being splitting each script up. If anyone has made 5000+ glyph fonts and has a better solution, please let us know. And if there are further question, please do email us directly.

The P22 blog post was an honest explanation without going into excruciatingly boring detail.

Thank you Miguel for not sugar coating it :)

Nick Shinn's picture

Well pardon me for sending you into DOUBLEBANG CAPITALS MODE, Miguel!!
It's true I couldn't be bothered to check with the foundry, but it was an instantaneous post on my part, to demonstrate that a guess could just as easily be sympathetic to the foundry as cynically accusatory.

Miguel Sousa's picture

That's the only way I have to react to uninformed people and comments based on assumptions. I use Typophile as a vehicle to share (give and receive) information, not gossip. People that want to output their BS are better served by itubemyconspiracytheories.com

canderson's picture

I've had to troubleshoot a lot of font related computer problems, and I can say that platform level bugs are the worst. This is because one by necessity must assume that the probability of their existence is small or everyone would know about it. In the case of the premier foundries, the actual specifications and feature support are being tested at the OS level. I sometimes define whether a font is "good" or "functional" based on whether it works in basic applications like Notepad.exe or TextEdit.app. So, it's not insane to be skeptical about OS bug complaints. It's also possible that the "bug" lies in some nebulous region that is better described as "compatibility". If only a couple fonts fail, the platform vendor could simply describe the problem as a lack of support for a particular feature. BTW, P22 London Underground, Arno Pro, and Mac OS are all things I would like to use if possible.

guifa's picture

The end price of the full font system is the same as originally planned. If someone wants just one script, they will actually be paying LESS
I did mention this, that this does benefit a user who would only use one of the three scripts, because otherwise, he would be paying for 200% extra glyphs that he would never use.

If I'm understanding the bug though, is it so much an OS X bug or Fontbook? Do third party font managers handle it okay?

Now if Apple would just hurry and CoreText to us...
«El futuro es una línea tan fina que apenas nos damos cuenta de pintarla nosotros mismos». (La Luz Oscura, por Javier Guerrero)

Miss Tiffany's picture

It is an OS X bug. When Arno was being beta tested the computers were different and that included different font management. The bug was acknowledged by Apple.

Christoph's picture

Thank you guys, that's pretty interesting.
I could not imagine why P22 spends much time into dividing the font, but I did not work with fonts that big yet. OpenType is such a nice thing in theory! ;)

kegler's picture

I could not imagine why P22 spends much time into dividing the font

Not for shits and giggles, that's for sure.

Christoph's picture

That's why I asked.

Nick Shinn's picture

Would a 1500-glyph font encounter problems?

paul d hunt's picture

Would a 1500-glyph font encounter problems?

it all has to do with a portion of the font file that lists all the glyph names, if this list is over a certain size, Mac OS will encounter problems including crashing of the system. The only way to test to see if you're in the safe range (that i know of) is to build (or test building) your font with the AFDKO tools.
Unless you're using really wacky long names, i think 1500 should be safe.

Miguel Sousa's picture

If I recall correctly, this bug only affects name-keyed OpenType CFF fonts in Mac OSX 10.4.8 and earlier. (CID-keyed OT CFF fonts go through a different "pipeline", I think)

Both MakeOTF and CompareFamily — available in the AFDKO — have a test to assess the fonts and will output an error/warning.

To calculate it "by hand", do the following:

1. Open Terminal (Mac) or Command Prompt (Windows), and type
tx -dcf –1 My_Font.otf > output.txt
tx is also available in the AFDKO. The kit contains the document CommandLineHowTo.pdf as well.

2. Open "output.txt" in a text editor and search for "### String INDEX". You should find something like this
### String INDEX (000012b3-000074f2)
count =2823

3. Startup Python and type (Use your own values; I'm using the values above)
2*2823 + (0x74f2 - 0x12b3)

4. A number will be output — in the case above is 30797. If this number is equal or greater than 32768 (32K), then the font will trigger the bug.

In sum, the problem happens if
(2*count + String Index table size) >= 32768
One way of reducing the number is by shortening the glyph names*. Other strings such as the font's Full Name and the Copyright may also contribute to the size of this number.

*Keep in mind that glyph naming rules still apply. In Arno Pro, we managed to get away from the problem by using uniXXXX names for most base glyphs, and uXXXX.x for most alternate glyphs.

Thomas Phinney's picture

Miguel is correct that CID-keyed fonts avoid the problem. Specifically, because they have no glyph names (they are "CID-keyed" rather than "name-keyed").

In our fonts, we ran into the problem for the first time right around 3000 glyphs. It would be an odd font with 2000 or fewer glyphs that would trigger the problem.

Interestingly, it turned out we'd come *really* close several times before, and just hadn't quite gone over the OS X string limit prior to Arno. It was really hard to debug, too, because there was nothing wrong with the font and the trigger was not at all obvious. We did eventually figure out it was at the OS level and Apple helped sort out the exact issue.

Cheers,

T

Syndicate content Syndicate content