Not sure about your own kerning but AFAIK Arno Pro uses an extension lookup type for kerning. Not every app can deal with this. Maybe XP7 is one of them? Do you have a chance to test XP8?

I'd like to hear more about this. I hope it isn't a matter of Adobe finding a way to prevent Quark from doing kerning -- rather analogous to the well-documented and never fully repaired fault that ruined ATM Deluxe on the Mac: it declared most of the Monotype and Bitstream libraries to be 'broken' when they weren't. (MacWEEK, 17-MAR-97, p. 1) It's not pleasant to have to report, but Adobe knew what it was doing. They desperately wanted MT and Bitstream to be declared 'broken'. Why else did they never fix the problem? And did they really expect reviewers to accept that these long-established libraries were broken? Yes, they did.

> I hope it isn’t a matter of Adobe finding a way to prevent Quark from doing kerning

For some reason that's the first thing that came to my mind: Seinfeld — The Mom and Pop Store.

JERRY: Hey, so where's my sneakers?

KRAMER: That's what I wanna know.

JERRY: What do you mean?

KRAMER: Well, I saw Mom and Pop this morning, but when I went by the store on my way home? The place was empty. Everything is gone. Mom and Pop - vrooop - vanished.

JERRY: So all my sneakers are gone?

KRAMER: I'm afraid so. And that's just the tip of the iceberg. I've been asking around - they didn't even have any kids.

JERRY: Mom and Pop aren't even a Mom and Pop?!

KRAMER: It was all an act, Jerry. They conned us, and they scored, big time.

ELAINE (amused): So. Mom and Pop's plan was to move into the neighborhood...establish trust...for 48 years. And then, run off with Jerry's sneakers.

KRAMER: Apparently.


ELAINE: Alright, that's enough of this.


@ Bill and David,
We, the Adobe Type Group, definitely have better things to do with our time than to think of ways to stop the kerning from working in other apps. It works in TextEdit and Notepad, so there's no reason for it not to work in QuarkXPress.

Your concerns about Arno Pro's kerning have no ground. It was all laid down before, more than a year ago at http://typophile.com/node/34059 . So please, next time do your homework first and stop wasting my time with nonsense.

@ David,
Are you still planning to e-mail me the "whole problem" ? Haven't received anything yet...

> @ Bill and David,

Why me?

> We, the Adobe Type Group, definitely have better things to do with our time than to think of ways to stop the kerning from working in other apps.

We tried to say that to Bill. Read the lines & between the lines, the whole so-called "conspiracy theory". So, next time jokes + an explanation ? :^)

I just wanted to know why OT kerning doesn't show in Quark 7. Not necessarily Adobe fonts - any OT fonts. Have you tried Arno Pro in QX7 Miguel?

I had thought there was something wrong with my typeface, so I just picked Arno Pro for comparison purposes to see if the kerning worked in different programs.

Nick Cooke

OK, I've read the other thread. It is just as I feared. Adam Twardoch states,

'The kerning lookup(s) in Arno Pro are so complex that the decompiler in FontLab Studio fails to interpret the code for it. But of course kerning is in there.'

It is just as I feared. Adobe has broken Fontlab. Again.

Now I cannot remember Yuri or Adam ever whining to me about Adobe giving them adequate support. That's not the Russian way, that's not the Polish way.

But Macromedia whined! Adobe never gave Fog adequate MM support. Application developers whined by the hundreds: Adobe never gave them adequate MM support. The type group is so paranoid and so arrogant that it never even gave Adobe's own app developers adequate MM support, which is why it took Apple and Lari Software, together, to provide adequate MM support, briefly, around the mid-90s. And why is it so few programs -- certainly neither Fog nor Flag -- still -- can do intermediate masters?

And when ATM Deluxe for Mac came out, and the Monotype and Bitstream libraries -- which up until then had worked impeccably -- were pronounced 'broken', Bitstream and Monotype didn't whine, they howled. Especially when Adobe didn't fix the problem. Even Microsoft has been bruised by Adobe's intransigence.

Why? Because Adobe at the gut level believes in proprietary standards.

John Warnock cried in public when he was forced to announce Type 1 would no longer be private -- soon after TrueType was announced. Nothing has changed since then.

The Adobe type group historically does everything it can to keep as much type technology private as it can, whilst guiltily promoting its open-ness at every opportunity.

Nick, I have tried Arno in Quark 7.31 and can replicate your problem. But I don't see the problem with any other of a dozen OT fonts I have tried. It looks like it's Arno that broke Fontlab and Quark.

Typical! How many designers said they'd never print with MMs because their service bureaus couldn't output the type? Adobe always said, 'old driver, old rasterizer. It's not our problem.' Like hell it's not your problem . . .

Somehow Apple managed to get all this right in GX in 1995 without breaking anything. Type really does need to be more integrated with the OS.

By the way, Arno kerning also is broken in Word 2007. And I can't replicate the problem with any other font. How much money would you like to bet that Adobe will offer a fixed, free, download, within a few days? No, it will take months and years for the applications to get it right. 'Not our problem' - - and yet more proprietary typefaces from Adobe.

Don't these guys even test their fonts? Like the rest of us?

Oh yeah -- of course they test -- in CS. Which has its own rasterizer. Of course.

WHY does Adobe build a rasterizer into its own apps? So it can use fonts other apps can't.

Is there something wrong with this picture ?

Right. So Arno Pro has had this problem for a year. My version of Word 2007 is totally up to date. When is Adobe going to fix the buggy product? Not that I like it, but there's an important principle at stake here. Anyone tried it in Quark 8? I'll try that in a day or two. If it doesn't work, I'll try to learn what Quark's position is -- though I should imagine it was something like 'Adobe should build fonts that work with contemporary products.' Imagine if MS tried something like this? But it would never happen. I guess the reason the font has never been fixed is that it just isn't important enough. But .... keeping it broken is just the way you keep lots of people from using it. Why don't they see it that way?

OK, I've tried Quark 8 on Vista; I'll try it on Mac later. Same thing: Arno doesn't kern; everything else does. Didn't notice many T kerns in MS-shipped FB Agency. David Berlow, if you're here, is that version of Agency without kerning?

Nick, may I suggest you rename this thread 'Kern problem in Arno Pro' ? The problem doesn't seem to have anything to do with Quark 7 or 8 per se, and doesn't seem to have anything to do with other OT fonts, and it does seem to be a problem with four apps so far: Fontlab, Quark 7 and 8, and MS Word 2007. No doubt there are many others.

So it's time to ask what Adobe QT found when they tested this font? If Adobe were now to claim that it didn't test the font on these apps, it would be difficult to believe. So then you have to ask, did Adobe know the font wasn't going to work and then decide to go ahead with it anyway, knowing that it would only be supported by CS?

What's really going on here?

Bill, extension format lookups are part of the OpenType spec and have been for a long time. If Quark or anyone else doesn't support them, that's an app bug or limitation not a problem with the fonts or any kind of conspiracy. The extension format is necessary when lookups get so big that they overflow the 16bit offset limit. While Arno might be unique among Adobe fonts in having an extension GPOS kern lookup, such lookups are far from uncommon in complex script fonts. I use them regularly. They are not only supported in Adobe apps but also by all Microsoft's OT Layout engines and, I suspect, by the various open source layout engines (ICU, Pango etc.). If it doesn't work in Quark apps, that's evidence that Quark is still lagging when it comes to OpenType layout.

Word 2007 does not support any GPOS kerning for either TTF or CFF fonts, so for CFF fonts it relies on a blackbox on-the-fly conversion of a subset of kerning, which is now several years old. It is possible that this conversion doesn't know how to handle extension lookups. Again, this is not a bug in the font but a limitation in an older piece of software (and more generally a limitation of Word in failing to support GPOS kerning).

With regard to problems with FontLab being unable to decompile some aspects of OTL tables -- and kern extension lookups are hardly unique in this regard -- this is due to a combination of factors. It is important to realise, and remember, that FontLab probably never decompiles OTL tables to a source that corresponds one-to-one with the original source for the font, even if that original source was also a FontLab file. What FontLab tries to do is to decompile the tables in a way that provides, within the limitations of the version of AFDKO syntax that they use, equivalent results, i.e. what you get is reverse engineered output not reconstruction of the original input. If a font is pretty simple, FontLab can do a reasonable job at this and you might end up with something that you can recompile. I've never had any luck with this, but then my fonts tend not to be simple. Also bear in mind that FontLab still uses an earlier version of the AFKDO code for OTL; I presume that Arno was made using either the current AFKDO or that Adobe needed to update their internal tools in order to be able to produce this font.

VOLT has had an option to use extension lookups since v1.2 (January 2006), and also to use PairPosFormat2, which is another thing that may not be supported by FontLab/AFDKO either in terms of decompiling or compiling.

So 'what is really going on here' is that Adobe have made a font following the published font format specification. So rather than saying 'Adobe should build fonts that work with contemporary products', I would say that contemporary products should support contemporary text layout, especially when that layout is part of an ISO standard specification.

Didn’t notice many T kerns in MS-shipped FB Agency. David Berlow, if you’re here, is that version of Agency without kerning?

AGENCYR.TTF has no kern table and no OpenType Layout tables, so no kerning. The absence of OTL is a pity, since the font does contain the variant A K M R U V W X and Y glyphs, but no way to access them (other than as raw GIDs via Adobe glyph palettes).

>So ’what is really going on here’ is that Adobe have made a font following the published font format specification. So rather than saying ’Adobe should build fonts that work with contemporary products’, I would say that contemporary products should support contemporary text layout, especially when that layout is part of an ISO standard specification.<

John! Exquisitely reasoned, but fatally wrong. We're talking about fonts here, and if a font doesn't work with just about everything out there, it's broken. If it doesn't work with Word, it's broke. If it doesn't work with Quark, it's broke.

Adobe applied precisely the same reasoning that you have here to its defense of ATM Deluxe for Mac. They said, basically, the entire Monotype and Bitstream libraries 'don't follow the published font specification.'

So what happened? The fonts are still with us -- unchanged, still working. ATM Deluxe for Mac is gone.

Look, I appreciate your viewpoint: in an ideal world, apps improve to meet new demands.

But your approach is way too ivory tower for this world. A font's just a font. It's got to work with apps that people are using every day that might be a decade old. Maybe one day the apps will be updated. But the user base never will be. And it's for users we build fonts.

That's why we test 'em. And if they don't work with Word, we don't publish 'em until they do work.

Seriously, what would happen if Microsoft released a font that didn't work with Word and Quark?

This is the daffiest manufacturing situation I have yet encountered with an Adobe font. I guess David Lemon isn't doing testing anymore. Surely he'd never allow something like this. Tell me it ain't so!

Couple more points:

1. A major goal of companies like MS and Adobe -- and also font developers -- is to prevent support calls

2. "If you need more than 200 kerning pairs, your spacing is bad." -- Wise words from Alex Kaczun.

3. If display problems are this bad already, think what printing problems are going to be like -- in the real world.

You have to wonder where the OT people's heads are -- GX had a foolproof mechanism for dealing with all of this -- everything fancy happened before the stream hit the PS print driver. Beautiful! Why can't they think like that today?

Seriously, what would happen if Microsoft released a font that didn’t work with Word and Quark?

Microsoft have released fonts that don't work in Word, and when they realise this they sit down and figure out whether it is a font bug or an application bug. And then they log the bug on their bug tracking system and either it gets addressed by the relevant group (MST or Office) -- perhaps not immediately, perhaps only in the next major system or app release -- or a decision is made to live with the bug if there are concerns about backwards compatibility or document reflow. Over the past few years, my font work has resulted in the overhaul of the entire Uniscribe Hebrew layout engine, and the acknowledgement and/or fixing of numerous bugs in Word's kerning support, and that's just the stuff I'm allowed to talk about. Over on the Typophile Arabic SIG page there is a report of a major bug in WPF's Arabic mark-to-ligature handling, which this morning is being investigated by WPF developers at MS. And what is consistently the case in all these is that the font has been made according to the specification, and the app or layout engine is doing something wrong.

You need to get your head out of the Latin-centric graphic design world in which a 'font's just a font' that is expected to be compromised to accommodate limitations or bugs in application support. To Microsoft a font is an enabling technology for users to compute in their own languages, and hence a key to new markets. This is why MST is part of the MS internationalisation group. The work MST does supports font use across the whole range of MS products, but their core business is script and language support.

I suppose if the only font you ever encountered with a GPOS extension lookup were Arno Pro, it might seem like a big deal. But as I said extension lookups have been part of the OpenType specification for a long time and they are used in a lot of other fonts. They just happen to be fonts that you don't encounter and that Quark doesn't even pretend to support. This is not some new or unknown technology that Adobe have capriciously decided to use in one font family, it is a well documented and reasonably common aspect of OpenType for the better part of a decade.

Quark has a history, during this same period, of first not grasping the importance of multibyte character encoding (I was at the Unicode conference when they announced that there were sticking to 8-bit codepages) and then playing catch-up by implementing a virtual clone of InDesign 1.0 OTL support and never pushing much beyond that. I presume they test only with relatively simple Latin OT fonts, have not encountered any fonts with extension lookups, and presumed that this was something they could postpone supporting. Actually, now I think about it, how does Quark do kerning? -- are they actually supporting GPOS kerning at all or are they, like Word, doing an on-the-fly conversion to old-style kern pair data?

Re. Alex Kaczun's comment about 200 kern pairs:

He was presumably talking about an 8-bit set of <256 glyphs (219 in the standard Windows Latin 1 codepage). Kerning is factorial, since adjacency means that there is a potential need for kerning between any unique pair of glyphs. This means that the potential number of kern pairs in a font increases exponentially with an increase in the number of glyphs: n!

Arno Pro Regular contains 2,845 glyphs. Now, obviously not every glyph is expected to occur adjacent to every other glyph, nor is the maker of the font likely to consider every problematic adjacency to be a high priority. So let's consider Kaczun's proposed 200 kern per c.219 glyphs, and how that scales to 2,845 glyphs. The actual potential number of kern pairs in 219 glyphs -- 219! -- is

1.038118334506655207907848120507e+419

[Remember that the exponent equals the number of decimal places the . needs to travel to the right (+) or left (-) in order to write the number out in full. So imagine the above number without the decimal point and with 389 zeros after it. Actually, no, don't imagine it. It looks like this:

103,811,833,450,665,520,790,784,812,050,700,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,00,00,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,00,00,00,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,00,00,00,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000]

Yes, that many. Which means that 200 kern pairs equals an infintesimal fraction of the total number of potential kern pairs:

1.9265626408095948382249858301366e-415 %

The factorial of 2845 is the staggeringly huge

2.6448619317498636060239040786471e+8593

If we presume a similarly limited and constrained set of kerning as a percentage of that number, we end up with

5.0954921878087836162237868848158e+8176 kern pairs

To put this in some kind of perspective, consider that the circumference of the earth is 18,933,146,496 picas. This means that, presuming an average glyph width of half an en (which is actually a bit narrower than the average width of glyphs in Arno Pro Regular), if you were to typeset the above kerning glyph pairs in 12pt type it would stretch around the world

2.691307643388966896537442227804e+8166 times

Oh, so that's not much easier to imagine. But that's the nature of factorial calculations, and that is why throwing around arbitrary numbers like 200 as the maximum number of kern pairs any font needs is silly. One thing that is very clear from the numbers above is that, as a percentage of the possible number of kern pairs relative to glyph set size, the kerning in Arno Pro Regular is much more limited and constrained than Kaczun's dictum.

I wonder what the expanded number of pair kerning in Arno Pro Regular is? To provide even reasonable kerning for a glyph set of more than a couple of hundred typical roman letter shapes requires class-based kerning technologies. Even so, you'd better be damn sure that the basic spacing is as good and efficient as it can be. I am working on kerning a display font with 4,517 glyphs, and even with strict limitations against cross-script kerning and use of contextual alternates to improve fit and avoid adjacency spacing problems, the number of expanded kern pairs is about 340,000.* But the number of class plus exception kerning pairs is less than 10,000, and I think I can probably whittle it down further.

*Which is pretty small considering that the potential number of pairs is
2.9936932664428451861273182257307e+14549

C'mon, John, you can play all the numbers games you want, but it doesn't alter the all-important fact that Adobe has introduced a font that only works in its own apps -- which all have their own built-in rasterizers.

This is of shattering significance.

Because it explains why Adobe is the only company to build rasterizers into its apps.

Not a good pointer to where things are going.

>You need to get your head out of the Latin-centric graphic design world in which a ’font’s just a font’ that is expected to be compromised to accommodate limitations or bugs in application support.<

I'm too busy waiting for you guys to solve the problems of A-z and 0-9. I realize that isn't where the money is, but it's what I'm interested in. Those are the simple things that get forgotten when you're worrying about 100,000 kern pairs. As you know, my only real interest is the Western European book -- I can't change that.

The other thing I'm slightly interested in is Adobe's historical attraction to proprietary font technology, which Arno Pro effectively demonstrates, in its small way.

>how does Quark do kerning? <

In Prefs/Project/General there's a switch to turn OT kerning on and off. I've asked Quark to look at this thread. I'm sure they would be delighted to send you an eval of 8, if you have time to look at it.

PS: Three great modern fonts designed to work without any kerning at all:

Trinite
Cheneau
Georgia

What I mean to say is, kerning is not important enough for you to break a font for it.

By the way, John, if you have 340,000 kerning pairs to deal with, I hope you're considering Adobe's advice that kerning pairs must be evaluated at 2400 dpi typesetter output - - no onscreen fiddling! Even so, they're always making errors . . . . in 250-glyph fonts.

> I wonder what the expanded number of pair kerning in Arno Pro Regular is?

Flattening the 1,2302 singleton+exceptions+class pairs yields 1,001,048 pairs. The italic fonts will have more.

> Adobe has introduced a font that only works in its own apps

Not true.

> Adobe’s historical attraction to proprietary font technology, which Arno Pro effectively demonstrates

Not true.

So

[not true] = [bugfix] ?

Bill, you are getting very confused here. This has nothing to do with either rasterisers or printer drivers. It has to do with OpenType Layout engines, and Adobe is far from the only company that 'roll it's own' in this regard. If Quark are offering anything like cross-platform parity in layout behaviour, then you can be pretty sure that they are also using their own OTL engine and not relying on system resources.

Now, if Adobe is guilty of anything with regard to OTL engines, it is the same thing that everyone else is guilty of: partial implementation of the OT spec. Even Microsoft, who are so far ahead of everyone else in terms of what they support, don't handle everything that is in the spec, in their layout engines or in their tools: they just support a lot more than anyone else. My big complaint against Adobe, which I know they're tired of hearing -- and which I also know they are addressing --, is lack of support for one-to-many substitutions in their core text engines. I'm not paying very close attention to Quark, since they are not a significant entity in the market segment in which I operate, but obviously I would consider lack of support for extension lookup types to be an unwelcome limitation.

I'm sorry, but your comments about 'proprietary font technology' really don't stand up. This is an ISO standard, Bill. The extension lookup types have been publicly documented since at least 2002. If using these lookup types indicates an attraction to proprietary font technology, then I'm way more guilty than Adobe, since I've used these lookup types for longer and in more projects. And so, I suspect, have plenty of my colleagues dealing with complex scripts. [If you want to blame someone for the existence of extension lookups, I suggest you start with Apple, since they are the people who defined the entire sfnt font format around 16bit assumptions, and hence made such mechanisms necessary.]

***
This is one of those discussions in which I have said everything that I want to say, and in which I am only likely to repeat myself going forward. In concluding, I'll note that my perspective is anything but 'ivory tower'. I comment from the rock face, and I do consider it the proper business of app developers to build roads through behind me to the place where I am blasting. :)

John, good discussion, several points.

1. That OT is an ISO standard is risible, as you know. It's part of MPEG. That means exactly nothing, even to MPEG people.

2. You yourself have noted elsewhere that parts of the OT standard which don't work should be discarded.

3. A font, no matter how complex, is never perceived by the vast majority of users as anything but a tiny functional widget that always works and never causes problems.

4. For Adobe to release a font that only works with its own apps is like Sony releasing a CD that only plays on Sony players.

5. It is possible for Adobe to make Arno work with existing software. They have chosen not to do so.

6. It is thus clear that Adobe is trying to seek competitive advantage. The message is that if you want quality fonts going forward, you'll have to use Adobe fonts only. Unfortunately, the font isn't compelling enough, technologically or aesthetically, to make anyone care. But Adobe has always tried to push the message that _only_ its fonts give reliable, professional, output.

7. It is significant that Adobe has had to give this Trojan away. Of course. Nobody would actually buy it. To get a Trojan to work, you have to get the Trojan to spread. So you give it away.

Bill, you’re making a fool of yourself.
Your basic ‘point’ (#4) has been disproved at the very beginning of this anti-discussion. Apple TextEdit and Microsoft Notepad are not Adobe software.

Tell me Florian, since when is Notepad used to set type? Or TextEdit, for that matter? That said, Notepad does indeed support ArnoPro's kerning. But that's unimportant, unless you use Notepad for word processing or layout. Do you?

What's important is that Microsoft Word 2007 does not support Arno Pro. Neither does Quark 7 or 8. Neither does Microsoft Publisher 2007. How long is the list?

Now please look at Adobe's thuggish response to David's complaint, above:

'We, the Adobe Type Group, definitely have better things to do with our time ... no reason for it not to work in QuarkXPress ... Your concerns about Arno Pro’s kerning have no ground...stop wasting my time with nonsense....'

This is the rudest response I have ever seen to a customer complaint in my life. And it's completely untrue. Arno Pro's kerning does, indeed, not work. It isn't nonsense.

The font doesn't work in Quark and Word 2007 and Publisher, and neither, probably, in the majority of Windows applications that do kerning. More testing would reveal this - - or Adobe, which has obviously done the testing, could simply come clean and publish a list of applications which it knows are broken by Arno Pro.

Maybe it works in TextEdit -- if so, that's a tribute to the GX-derived OS support. Yet Apple isn't even a co-developer of OT. MS and Adobe are. And MS doesn't believe in putting the depth of font support at OS level that Apple does. It's a justifiable though very expensive position to take.

In any case, a major selling point of OT fonts is that they're cross-platform compatible. It looks like this one is only partially cross-platform compatible.

Finally, to those who take exception to the Trojan analogy, let me put it this way: if Arno were not a Trojan, Adobe would have said, upfront, when they released it, 'this font only works with apps x and y; it doesn't work with Quark, Microsoft Word, Microsoft Publisher', etc. etc.

When did Adobe provide any such advisory? Over to Adobe.

People talk about Quark's long-past rude customer support. What compares to

’We, the Adobe Type Group, definitely have better things to do with our time ... no reason for it not to work in QuarkXPress ... Your concerns about Arno Pro’s kerning have no ground...stop wasting my time with nonsense....’

I gave CS3 Mac Directory magazine's annual award this year. On balance I thought it was earned, mainly for the small though brilliant improvements in Photoshop and for the purchase of Dreamweaver. I didn't think much of what InDesign and Illustrator had racked up since CS2. But had I known that Adobe was bundling CS3 with core fonts that don't work with other apps, I would have placed the award elsewhere, explaining my reluctance. I only wish I'd been aware of this problem a few months ago.

And frankly, Adobe's customer response in this forum is enough to ban it from ever getting an award from me until the company can prove to me that it is treating its customers better than this. I'm going to cross-post this in the 'Quark 8 doesn't suck' thread.

One more point:

Adobe advertises OT as a _replacement technology_ for MM ... not a good way to advertise it, but it's how Adobe has chosen to present OT.

One of the problems with MM fonts was that there were endless complaints. How many times did one hear, 'I will never output with Kepler again'; 'I will never output with Jenson again'; 'I will never output with Minion again'. One endlessly heard of print disasters.

Adobe blamed it all on three things: out of date print drivers, out of date versions of ATM, out of date or otherwise non-standard PS rasterizers.

Like John, in his defense of Adobe, I bought into all of this for years, until I realized that like an incurable alcoholic, Adobe was always blaming everybody but itself.

What Adobe cannot afford to do now is to give OT the same kind of reputation that MM earned.

A well-known type designer on this list who often teaches graphic design students told me that he always asks his students if they have ever heard of multiple masters. None have. Then he asks how many have heard of OpenType. Less than 10% have. OT mustn't go down the same path.

Nick,
Any news on your font's kerning?

You wrote,
I just wanted to know why OT kerning doesn’t show in Quark 7. Not necessarily Adobe fonts - any OT fonts.
Is it really happening with all OT fonts?

I guess the customer left. And who can blame him?

However, for the record, further research has revealed that although Arno seems to be unable to work properly in Quark 7, it does work properly in Quark 8. However, if the file is a legacy file, it needs to be forced into Quark 8 native format. I'll post further details if anyone is interested.

That leaves all the significant apps of OpenType's co-developer, Microsoft. It would be useful if others could test other applications on both platforms with this troublesome font.

Bundling it with CS was not something I can approve of.

In the meantime, I'm constantly delighted at how lovely this site looks in Georgia -- a font designed to work well without any kerning at all.

And a final (I hope) point. Fussy book designers often turn kerning off even in old Adobe Originals fonts. One reason is too many annoying kerns around punctuation which can cause run-on sentences and other aggravating problems. Other fonts have this problem. For example, if you are publishing for the UK and using single quotation marks rather than double quotation marks, you'd need to turn off kerning in the lovely font Scala.

Sometimes less is more !

I'll just note that any kerning that works in TextEdit almost certainly works in other Apple apps such as Pages and Keynote.

(The other ravings in this thread aren't worth anyone's time to comment on.)

T

Thomas from Adobe:

>(The other ravings in this thread aren’t worth anyone’s time to comment on.)<

Unless you are a user of Microsoft Word or Microsoft Publisher or a thousand other applications.

What a dreadful way to treat one's development partner -- and its apps.

And it is typical of Thomas to say, 'almost certainly works'. That translates as 'too lazy to test; would rather blog.'

With tech support like this, who needs enemies?

Altogether, we are not getting a good impression either of Adobe's testing or of its manners.

We're also not getting a good impression of Adobe as a team player with Microsoft in the development of OpenType. And I used to think that Microsoft had an arrogant corporate mindset. They don't, really.

Adobe's Thomas Phinney carefully noted that Arno 'Pro'

'almost certainly works in other Apple apps such as Pages and Keynote.'

I misread Apple apps to mean Mac apps. I don't think he means 'Mac apps' and I think he knows a lot more than he's telling. For example, I can't even get Arno to show up in the font menu of Word:Mac.

How many copies of CS was Arno Pro bundled with? On how many millions of systems was this font automatically installed, without warning users that this font does not work with 99% of their apps, Windows or Apple?

Folks, this has never been done before in the entire history of software distribution. Nobody has ever released a font on this vast scale that didn't work with 99% of existing software.

Then, when one user -- out of a gazillion -- has the temerity to complain in a public forum, Adobe responds:

’We, the Adobe Type Group, definitely have better things to do with our time ... stop wasting my time with nonsense.’

Could this be Adobe's idea of PostScript evangelism? Customer support? What's happened to the people who were so good at getting Adobe's message across a decade ago, like David Lemon, Rick Wulff, Chris Holm, and a hundred gentle PR giants?

Footnote: a specific, and very common, example of Adobe-style kerning problems from a Kathleen Tinkel article back in 1998:

"... if you have a sentence whose last word ends in y followed by one that begins with W, the word space is
reduced to roughly 10 percent of its normal width."

(because of too-aggressive kerning between y-period and period-W)

Kathleen also published about the long-standing problem where ATM Deluxe (Mac only) maliciously reported that 600-odd fonts in the Monotype Classic Library were damaged. Monotype went ballistic, noting that the fonts had been built with Adobe's own tools. Then Adobe said they didn't think Monotype had properly paid the licensing.

Somehow this was all allowed to play out through five or six dot releases of ATM Deluxe, until it became a null issue with the adoption of OS X and the vanishing of ATM.

The kerfuffle with Arno reminds me of this ancient history.

By the way, ATM also buried an enormous amount of TrueType (including all of GX) and any MM font generated by Fontographer. All this time Adobe was talking about how open they were in supporting independent MM development .... no secrets .... yeah

>That translates as ’too lazy to test; would rather blog.’
>I can’t even get Arno to show up in the font menu of Word:Mac.

All 32 Arno Pro fonts work for me on Word (MacBook Pro, MS Office 2004). However I checked a "Ta" pair and the kerning was indeed whacked.

Miguel Sousa has already referred to this topic's previous coverage:
http://typophile.com/node/34059
http://www.typophile.com/node/39130

It's seemingly headed here:
http://

>Miguel Sousa has already referred to this topic’s previous coverage:

Strange, then, that none of those prior threads mention all the other software, Mac and Windows, that Arno 'Pro' doesn't work with.

Here we have a font that gets distributed free to x hundred million users, and only works with 1% or less of the software that is out there. All of this in the name of kerning, some of it demonstrably bad. What are the overwhelming technological and aesthetic reasons that this font needs to be so messy?

And given that it is so messy, and doesn't work with 99% of software, why distribute it on such an unprecedented scale? Why not simply offer it to the specialists who might need it. Are there any?

I'm failing to see the reason why any of this should be happening.

Just tell me why it has to be like this?

And who needs Adobe kerning anyway, with examples like Kathleen's and the A, left single quote, right quote, comma example that Charles pointed out in an earlier thread.

Adobe's Sousa responds that they'll 'log' these problems.

Folks, people have been complaining, vigorously, about these Adobe kerning problems for nearly 20 years.

Do you really care if your problems are 'logged' or not? They're never, never, never, going to be fixed. After nearly two decades of customer complaints, I think we can safely make this statement.

Adobe is always right; the customer is always wrong. When was the last time Adobe fixed a font? Offered a new version as a bug fix?

To illustrate that Adobe-style kerning still has not progressed after twenty years of customer complaints, here's Arno Pro Display in Quark 8:

Why is it you don't get problems like this in Carter & Cone fonts, or Font Bureau fonts?

And not to add insult to injury, but the spacing doesn't look very good either.

Why is it you don’t get problems like this in Carter & Cone fonts, or Font Bureau fonts?

If this is a problem of commission, you get other related problems in their fonts (and mine), of omission.
For instance, if A_rightquote is not kerned tightly as per Arno, an all-cap setting with apostrophe is gappy.
In an all-cap setting, you're much more likely to have an apostrophe than a period--isn't that the logic of Robert Slimbach's kerning strategy?
Similarly, because he fits his letters closely, he doesn't like to see things suddenly open up when a couple of points are set side by side. Hence the vigorous kerning between quote marks and periods/commas.
In general, the Slimbach fit style produces very slick settings, but it does have the odd glitch (which will I assume soon be solved by triple character kerning, but only supported in InDesign :-)

I haven't commented in a while, but Bill's ranting managed to bring me out. I have only this to say... QXP, Word, and Publisher are not the right apps for testing the quality of your fonts. if you are a typophile, chances are you prefer InDesign ... or perhaps Tex.

Bill, unwelcome results in kerning are obviously a different issue from lookup format support, so I'm not going to dilute this thread by getting on a tangent about different approaches to triplet spacing issues such as you illustrate.

But since, as you report, Quark 8 appears to be supporting the GPOS extension lookup type, I feel a certain vindication of my previous insistence that this is a layout engine implementation issue and not a font issue. Obviously someone at Quark thought so to. I worry that perhaps you still don't realise just what the extension lookup type is, and are misinterpreting the whole situation as a result. The extension lookup mechanism is not some complex new way of performing a glyph substitution or positioning operation: it's a 32bit offset locator. In other words, it is just typical GPOS or GSUB lookup data stored in a different location with a pointer. This is bread and butter to any programmer, and supporting it is pretty trivial.

http://www.microsoft.com/typography/otspec/gpos.htm#EP

>[GPOS] is bread and butter to any programmer, and supporting it is pretty trivial.

Not for Microsoft it isn't. Otherwise we'd have had it in Word and Publisher long ago.

It's been Microsoft's spec for 11 years. They've had enough time to implement it and they haven't, for what are doubtless good reasons to them.

John, I appreciate your viewpoint -- you've contributed enormously to OpenType and you're an important evangelist for it. But I still fail to see any justification for what Adobe has done here, either technologically, or aesthetically. (And Adobe's response is most unsatisfactory. Was David Lemon ever so rude?)

I can, however, see it as a marketing move towards effectively proprietary type formats.

Adobe's canny blurb for Arno here

http://www.adobe.com/type/browser/landing/arno/arno.html

states,

'In addition, all glyphs in an OpenType font can be accessed in Adobe InDesign® software, whether or not they are covered in a layout feature. OpenType fonts, coupled with the advanced typographic control offered by a program such as InDesign, also let type-users take advantage of advanced justification, optical margin alignment, hanging punctuation, and optically sized masters ....'

The implication is that only InDesign will work with this font ....

And as you know, Adobe could perfectly well have implemented this font so that it would work perfectly with all extant software.

So again: why?

Let me put it this way. If you were the program manager for a piece of software that was going to shipped onto several million systems, each containing a huge variety of software old and new, would you deliberately bundle a font that didn't work with most users' software? Why don't you ask someone like Lori Birtley at Microsoft if she'd let this happen? (Last I knew, she was program manager for Publisher; years ago she was program manager for PageMaker, and she certainly was one of those who started the K2 project that grew into InDesign.) I'd like to hear her thinking on this.

> Now please look at Adobe’s thuggish response to David’s complaint, above:

Now who said that wasn't my mistake? — " For some reason that's the first thing that came to my mind:" instead of "this is"? . Who said that all of my jokes are clear & fine? And why do you think that other people see it as "complaint"?

The bottom line is: Alright, that’s enough of this!!!

I emailed Miguel; he emailed me back. We are fine and happy.

So, please enough of this!!!

Bill, I'm afraid you really don't understand the technical issues involved here.

Not for Microsoft it isn’t. Otherwise we’d have had it in Word and Publisher long ago.

Every single Microsoft OTL layout engine supports extension format lookups. I have made plenty of fonts that use this format, and they all work fine in Word. The issue for Word with Arno Pro kerning is not the extension format lookup per se, which it would have no difficulty supporting for kerning in any Arabic, Hebrew, Devanagari, Thai or other complex script font. The issue for Word is twofold:

1. Word does not employ any OpenType kerning at all for non-complex scripts. For TTF fonts, this means that there is no kerning at all for the Latin script unless that kerning is included as old-fashioned kern table pair kerning (either in addition to or instead of OTL GPOS kerning in the font). For CFF (PS OT) fonts, since these are by design not supposed to include kern tables, Word -- like various other Windows apps -- must rely on on-the-fly conversion of GPOS to virtual kern table kerning. Since it is risky to present any application with fully expanded class kerning, since many have undocumented limits in kerning support, this conversion involves subsetting kerning for specific 8-bit codepages.

2. Such kern data conversion is performed in Windows, as I understand it, by a quite old piece of code in a blackbox mechanism provided by Adobe (probably originally written for ATM). Because this code was presumably written before the OT extension lookup format was defined and added to the font format specification, it not surprisingly doesn't know how to handle Arno Pro. That, at least, is my analysis of the likely cause of the kerning failure in Word.

Bill, I frankly don't much care if you want to chastise Adobe for deciding to ship a font that uses a mechanism that is not universally supported in current applications. It seems churlish to me, because pretty much every font I ship only works fully in a small number of current apps, and I expect fonts and apps to develop in tandem, but I know that I'm unlikely to dissuade you, after all these years. But I do care that in your crusade you don't misrepresent or muddle technical facts, because people actually rely on technical information to make a living and form decisions about time and resources. If you want to attempt to deduce reasons -- however unlikely they may seem to me -- why Adobe might have decided to use a particular technology in a particular font, go ahead. My concern is with the accurate representation of that technology.

.

>My concern is with the accurate representation of that technology.

Microsoft will be delighted to have this information so they can finally go and fix Word and Publisher.

No but seriously, John, the issue isn't technical. So you don't have to be condescending about the technical issues here, least of all to me who despite my amateur pose cut my teeth on 8080 assembler.

The issue isn't how to do it but why Microsoft isn't doing it. They obviously don't think it's important to do. Who are we, in the font industry -- the the least significant of the bottom feeders -- to argue with Microsoft's priorities?

So: why did Adobe implement kerning in an exclusive manner when it was perfectly possible to implement it in an inclusive manner?

>pretty much every font I ship only works fully in a small number of current apps, and I expect fonts and apps to develop in tandem<

That's a luxury most of us cannot dream of. And I doubt you ship broken fonts to several million customers, free. If you did, you'd soon be out of a job.

In any case I am philosophically opposed to it. I expect everything on my computer to work and keep on working. I don't expect a font to break it. That just makes people queasy about fonts. And about OT in particular.

Oh - - it just hit me - - surely - - it couldn't have been - - you - - who talked Adobe into doing this thing? How daft! Gosh -- did you actually do the work on the font? Well, I'm glad if you did, for the money, but it was a very silly thing to do just the same. Best to keep quiet about it, I should say. Nevertheless, bravo, if so!

So.... The answer to my original question is "I don't know". ?

Nick Cooke

Sorry Nick, we got a little carried away with philosophical concerns here. We do have the answers. The kerning definitely works in 8. I'm not sure about 7 but I can't find any way of making it work in 7.

Here's what Quark says assuming you are using 8:

We took a close look at the kerning differences we see in the screen
shot of the differences between QXP 8 and Indesign. When we created a
new project in QXP 8 and ensured "Use OpenType Kerning" was ON, using
the same Arnold Pro Regular font at 48 pt., the kerning relationships in
QuarkXPress 8 were exactly the same as in InDesign. However, when we
turned OFF Use OpenType Kerning, we ended up with the same results as
the link shows. This tells us that QuarkXPress 8 is working as designed
and is correctly reading the built-in kerning tables for the OpenType
font.

This leads to one of two possibilities:

If this behavior is occurring in a brand new project created from
scratch, and Use OpenType Kerning is showing as ON in the QuarkXPress
Preferences, then the XPress Preferences are damaged. Essentially what
often happens with Preference-level damage is that a setting will show
as ON when it is actually OFF. The remedy is to quit out of QuarkXPress
8 and create a new folder in the QuarkXPress 8 folder, naming it
"Preferences". Then create a new project and ensure Use OpenType Kerning
is ON. Then see what happens with the Arnold Pro Regular font-It will
likely show the correct OpenType kerning.

If it is occurring in a legacy file, i.e., one that was created in
version 6.5 or earlier, these versions did not support OpenType kerning,
and, in order to preserve the text flow, QuarkXPress 8 will by default
for legacy files like this turn Use OpenType Kerning OFF for that file
only. If you want to force the file to flow using OpenType Kerning, go
to File>Open. In the Open dialog highlight the file. Then hold down the
Control and Option keys and click open. This will rewrite the file into
QXP 8 text flow code (along with rewriting the entire file to QXP 8 file
structure), which should tell QXP to turn Use OpenType Kerning ON.

This confirms my experience. Any new project in Quark 8 will work fine. If you're bringing an old Arno document over from 7, though, you will have to ctrl-opt-click Open. (Mac)

I can't help noticing the difference between Quark's genial, informative tone and Adobe's

’We, the Adobe Type Group, definitely have better things to do with our time ... no reason for it not to work in QuarkXPress ... Your concerns about Arno Pro’s kerning have no ground...stop wasting my time with nonsense....’

In an earlier post here, Nick Shinn defends Adobe's kerning of A right quote comma, but here,

http://www.typophile.com/node/39130

Adobe's Sousa writes, with what I can only take to be superbly Freudian cynicism,

Please do. I’ll make sure these bugs get logged, and will certainly be fixed next time the fonts get revised. No need to post them here. You can send the material directly to me at msousa adobe.com

What would be nice is if you could post a list of unusual kerning combinations, like the quote/quote examples you’ve just provided. This type of thing should actually be posted on the Wiki. If you don’t know how to do that, just start a new thread.

Mark this - - Adobe promises these fonts will certainly be fixed the next time they get revised.

Ladies and Gentlemen, place your bets!

While we're waiting, perhaps we, too, can spend the next 20 years posting a list of unusual kerning examples on the Wiki. If you don't know how, just start a new thread!

Finally, Nick - - are you at the point with Quark 7 that you can't get any OT font to kern properly? If so, check, as described above, that OT Kern preferences are on; if fonts other than Arno still don't work, in new documents, then you may have to trash preferences, but for a definitive answer, it would be best to give Quark more information. Please let us know in detail, if you have the time, how you are getting on with this.

Just in the past couple of days, a paradoxical situation has arisen for me. I generated a very simple OT font for a designer which includes extra f-ligatures, nothing more. I followed Fontlab's instructions to the letter. The ligatures work in Quark 8, but not in InDesign CS3!!!! So here we have an instance where Quark's handling of OT is better than Adobe's.

I'm afraid that highlights the weakness of OpenType - - the standard is so broadly written (when I say broadly I mean badly but I am trying to be polite) that nobody can ever be sure which feature, implemented which way, will work in what program.

This after a decade of OT ? Akkkkkkkkkkkkkkkkkk!

My sympathies are with the MS app developers on this . . . . OT is so leaky it's like a moving target.

Bill, I am sure you approached Quark with the same tone of voice that you approached Adobe with in this thread. I am also sure that you are quoting Quark as manipulative and selective as you are quoting Adobe in this thread. I am even more sure that you turned what could have been a very informative thread into a childish namecalling competition.

Thanks.

I don't think you're right, Artur. Had we left it to Adobe, we'd just have Adobe's response, which is that there isn't and never has been a problem, either with its kerning system or with its OT implementations. And please note that Microsoft is not making incompatible fonts. This thread should never have existed, because the problem should never have arisen. A technical problem as anomalous as this does not arise for no reason at all. The method may be painful, but we have learned something about why it arose. My quote from Quark is verbatim. Mail me, and I'll mail you their note. Ten, fifteen years ago, you could go to Adobe with questions and get responses from high calibre people like David Lemon, Rick Wulff, Chris Holm, and many others. Where are they now?

In the meantime, would you like to take on any of the issues? The responses in this thread have been pathetic. Adobe's in total denial, Microsoft's not interested, John's reflex-evangelizing any threat to OT because his livelihood depends on it. Badly as I may be doing it, I am the only person representing the voice of the user, other than the user who started the thread. Where do you stand?

And please note that Microsoft is not making incompatible fonts.

Oh, right, like those Vista C-fonts, please give me just a quick reminder how I can use their OT layout features in Word again.

Or how to access the kern pairs from the second kern subtable in Windows.

I didn't realise that I'd be opening a whole conspiracy theory can of worms when I posted this thread. I just wanted the solution to a problem.

Well thanks for your help everybody ;^P. I've sorted it now and it all works. My fault I presume - I didn't change anything in QX7. I somehow made it work by messing about with FontLab.

This is a workhorse family I developed for wayfinding signage for the town of Morpeth in Northeast England. As you can see the kerning now works for the standard set and contextual alternates.

Nick Cooke

Ps Bill Troop, how do you ever get any work done? you seem to be writing lengthy essays all day long :^)

Jens, mea culpa - - I forgot about the C-fonts. Once through with them was enough for me -- and poor John! (And give him credit for this: he does fix his fonts when they need it.)

Nick -- this is hilarious! So glad you got everything solved. You're right I am posting too much. But I always do at least one glyph between posts. Lots of interesting things in your font! Dot of i seems high but I expect you have reasons for that, and I love what you're doing with the t and its variations.

I wrote: Oh, right, like those Vista C-fonts, please give me just a quick reminder how I can use their OT layout features in Word again.

Or how to access the kern pairs from the second kern subtable in Windows.

Just to clarify: I don't consider these fonts broken, nor Arno Pro, just because there are some systems or applications that can't make full use of them.

Where do you stand?

Bill, your last question seems illustrative for your approach to this; If you're not with us, you're against us...

The fact that a specific version of Quark doesn't support the kerning in Arno Pro doesn't tell me anything.
My guess is that we could have had a lot more respons from the guys who can shed some light this on if you hadn't been shouting form the tallest tree from the beginning, constantely manipulating the discussion by misquoting Miguel.

Please do continue asking questions, critizing decisions, but do it in a way that doesn't create this current ambiance, which in my opinion is totally out of place in a community like this.

Artur

>please give me just a quick reminder how I can use their OT layout features in Word again.

Jens, I brought up the issue of MS apps and OTL (lack of) on the OT list seven years ago! (October 2001) Microsoft's wonderful Greg Hitchcock then promised he would escalate it to the highest levels. One problem was that the app developers didn't want to support OTL because of perceived performance hits, which Greg took issue with. Whatever the case, seven long years later, we still don't have much OTL typography support in MS apps. This just shows that it takes a long, long, long time for people to do anything about fonts when it doesn't seem important to the app people. Actually, it doesn't. OT is very, very unsexy and most app developers are not interested in it. And most customers aren't interested in it. People on Typophile are interested in it. So what?

Isn't it funny that this thread would also never have occurred had Nick not chosen to use Arno as his example, since that seems to be the only Adobe font that doesn't work in Quark 7. In the meantime, I need to solve why my OT font for Nicola works in Quark 8 but not in InDesign CS3. This is still a format for geeks, not letter designers. I wish it weren't so, but at the rate we're going, it's obviously going to be a long, long time before things get straightened out.

By the way, Adobe's OT support has always been partial, and there were one or two features that Quark implemented in 6 of 7 that Adobe hadn't yet done -- I've forgotten which. There's something the matter here. App developers are being forced to do way too much work, and take too many hits, whatever John or anyone else says to the contrary. The co-developers of the format -- Adobe and Microsoft -- are both app makers and have substantial difficulty in supporting the format, 11, 12 years after it was specified. You can't get away from that. AND USERS ARE CONFUSED.

Artur, you may dislike my tone, but I deal in facts not tone. You say I misquoted Miguel. When I say someone has misquoted someone, I illustrate with the quote and with the misquote. I warmly recommend this method to you. It is infallible.

How pleased I am that someone told me about this thread! (I'm sure everyone else will be glad to see me reviving it.)

Bill, I want to call attention to this statement you made:

Folks, people have been complaining, vigorously, about these Adobe kerning problems for nearly 20 years.

Do you really care if your problems are ’logged’ or not? They’re never, never, never, going to be fixed. After nearly two decades of customer complaints, I think we can safely make this statement.

... When was the last time Adobe fixed a font?

I want to note that quite a lot of the kern pairs that you are complaining about (cited by Kathleen Tinkel ten years ago) were, in fact, fixed by me many years ago during the conversion to OpenType. I can't say I got all of them, but I was aware that there were many such kerns in various Adobe Originals, and I tried to remove those when they crossed my path. Furthermore, some kerning problems (among many, many other bugs) were recently fixed in the revision of the Adobe Type Library. Some of these had been logged from customer bug reports (a.k.a. complaints), and some were flagged internally. In any case, all these logged kerning bugs were considered during the revision and many were corrected. Unfortunately many of the more subjective, less-offensive kerns were left alone, because with any change that affects text flow, we are concerned about the impact on legacy documents. You are welcome to state your own preference that all kerns which you consider wrong be fixed immediately, but there are always matters of personal taste and concerns about backwards compatibility which fall into the mix.

Which leads me to the next hilarious quote:

I deal in facts not tone.

You've asserted more than once that we are not interested in customers or bugs. I think that if you asked around you would find many people who have had their Adobe font complaints acknowledged with great interest and appreciation. I can tell you that we in Adobe Type are keenly interested in knowing about, evaluating, and fixing bugs -- but our ability to do so rapidly and completely is always limited by time, resources, and other practical matters. I think for the font profession and for the software industry in general, this is not unusual.

Erm... sorry to hijack my own thread, but I regenerated the fonts over and over because I was still having the same problem. I tried selecting the type and changing to a number of completely different random fonts, then trying my new fonts - still the same problem. Then it occurred to me to try a new document to test the fonts. Bingo! they all worked perfectly.

Nick Cooke

Nick --

At the risk of re-igniting the fires, let me ask: Did you try Arno Pro in the new document also, and did it work then?

-- K.

Moving on from the specific kerning and application issues...

The problem with fixes or upgrades to fonts is how to get them to the licensed users of previous versions. This isn't a problem only for fonts, but it is one which system and application developers have largely solved, using 'push' mechanisms to notify users via their Internet connection of updates, often with the option to download and install the updates. I have not seen such mechanisms used for font updates, and I suspect some important pieces of such a mechanism for fonts are currently missing. An OS or application can carry its own code for connecting to a server, checking for updates and alerting the user. For fonts, this would presumably need to be some kind of separate applet (or applets, presuming that font vendors would have their own: an Adobe Font Updater, a Monotype Font Updater, etc.). It gets messy in a number of different ways, especially for fonts that are available through cross-licensing from multiple vendors.

By-the-way who or how do I contact adobe regarding updates to their fonts.

The Poetica I have has some funky kerning.

Version 1.006;PS 001.003;Core 1.0.38;makeotf.lib1.6.5960

Thanks,

Mikey :-)

Kent, - yes I did, no it didn't.

My fonts worked though, that was my main concern.

Nick Cooke

Interesting. Thanks for the report, Nick. -- K.

The problem with fixes or upgrades to fonts is how to get them to the licensed users of previous versions.

I'd have said the problem was how to get foundries to offer fixes or upgrades in the first place.

John, software manufacturers usually deliver upgrades and bux fixes by a dedicated webpage.

When was the last time you saw Adobe offering a fix or upgrade for a font?

Yet Chris Slye writes,

I can tell you that we in Adobe Type are keenly interested in knowing about, evaluating, and fixing bugs — but our ability to do so rapidly and completely is always limited by time, resources, and other practical matters. I think for the font profession and for the software industry in general, this is not unusual.

Chris, can you point me to a single web page since the invention of the internet where Adobe offered a bug fix for a font?

Can you tell me of a single instance in its entire corporate history where Adobe ever sent out bug fixes to registered customers of a font? (Email costs nothing. What about before email? I can still remember Quark sending 3.3 or something like that to all registered users -- unsolicited -- via DHL. I don't remember Adobe ever doing that for any of its products, ever.)

Adobe also says,

Furthermore, some kerning problems (among many, many other bugs) were recently fixed in the revision of the Adobe Type Library.

Well, Chris and Adobe,

was a single customer offered a free upgrade for the bugs you belatedly acknowledge? (I use the word belatedly advisedly. This is the first time in 20 years I have heard Adobe acknowledge a font bug.) If so I'd like to interview that customer.

our ability to do so rapidly and completely is always limited by time, resources, and other practical matters

Fair enough. But twenty years?

Re the A-quoteright-period problem in Arno, is this something new? I don't see it in Minion MM, for example.

It almost seems that Adobe is occasionally (and without ever informing customers) 'fixing' kerning bugs only to add new ones.

In Arno we have the case of a font with so many kerns that a special, unsupported kerning technology has to be used which is not supported by the vast majority of programs.

Why not simply use fewer kerns?

Why build fonts that are not supported by most software?

You gotta wonder.

What I do know is that I have seen books from major publishers that show lavish and careful use of Minion and Minion expert, but where kerning is turned off because Adobe kerning causes more problems than it solves. This from designers who are going through the pain of using all f-ligs, small caps, osf, etc. where appropriate, in 700-page books.

Anyway, that's the thing. Designers are saying that Adobe kerning causes more problems than it solves. Maybe it's time to stop the gimmicks?

Designers are saying that Adobe kerning causes more problems than it solves.

I am somewhat conflicted about how many kerns to put in a font, and according to the typeface, have followed different kerning schema. These days, I'm reducing the amount of kerning, both for aesthetic and practical (huge glyph sets in OpenType fonts) reasons.
I don't think there's a simple answer.
The fact that some typographers turn off kerning is not necessarily an argument against it.
As far as Robert Slimbach's types go, I used Myriad Condensed a lot when I was doing editorial and brochure design, and I found the tight fit was brilliant.
However, I can see where those typographers who turn off kerning are coming from, in my own response to new versions of types such as Univers, which have the copious kerning their metal predecessors lacked. To me, Univers doesn't look like Univers any more, and I think it may be the metrics. In particular, kerning after the "r" does seem to effect the tonality.

Nick, that's a subtle point since in Lino Univers New there are only three lowercase kerns after r, each of -12: s, q, o. The kerns in Adobe Univers are quite different. I know some think Bitstream's version is the best. One thing I'm reasonably sure of is that most metal versions would probably appear quite a bit looser at 8 or 9 points than any of the comparable photo/digital versions. I don't know the font well enough to comment further.

I would like to make the point that extensive kerning does not necessarily dictate tight fit. Kerning can be strong or weak; can often, with advantage, be positive, not negative, depending on the scenario. It would be interesting to look at the Berthold Univers, which is probably optimized for a smaller size than the others.

Let's move on to a more interesting question: who on earth thought up Adobe's daffy kerning system to start with? We know that Sumner Stone was responsible for the early triumphs of the type group, including the Originals program. Yet looking through Cycles (which, along with Miller, I would exchange for the entire Adobe type library) I don't see any of these crackpot kerns that have caused so much trouble. For me, a Matthew Carter font is like a master class in type. Everything, not excluding the interesting kerning system, is a lesson that is just within my reach to grasp. But when I look at Cycles, I wonder if I will ever know what it's all about . . . . just look at the k and ask why he does what he does - - so eccentric, so willful, and yet it all makes sense. Sorry, I can't snip the k until I restart, and I can't restart until I've moved some files from my dying Vista drive to my Mac . . . . can someone else?

Bill, I can't recall that Adobe has ever replaced a customer's font for free due to a bug. I'm not sure it's never happened, but I don't know of any such case off the top of my head.

This is the first time in 20 years I have heard Adobe acknowledge a font bug.

Thomas talked about the library revision and its bug fixes at TypoTechnica in April 2007, and mentioned this talk on his blog the following month, so the bug fixes are not really the revelation you perceive them to be. I'm sure the revision has been mentioned elsewhere. (I just gave a presentation on it at TypeCon.)

I think you would agree that it's better to fix bugs than to leave them unfixed forever. Our first priority is to identify font bugs and log them, and then fix and release them. Getting them to customers is a distinct step. Obviously the latest fonts are available now to customers who want to buy them.

Regarding font "updates", this is something we're hoping to offer soon. The fonts available for upgrading will likely be quite limited in the beginning due to the complexities of font licensing and so on. My point here is that, like everyone else, we are quite interested in offering this to customers and we're working on it. It is not a trivial undertaking.

As I mentioned in my TypeCon talk, fonts are rapidly becoming bigger and more complex, such that they are practically like "little applications". We expect to continue to find, log, and fix bugs, like any application developer would.

>Obviously the latest fonts are available now to customers who want to buy them.

Christopher, people who buy fonts at all spend vastly more on fonts than they do on apps. I'm sure it's not uncommon for designers to spend ten or twenty or forty times as much on fonts as on all their other software and perhaps even hardware put together. And, as we're always telling people, fonts do take an enormous amount of work.

But fonts are priced accordingly. For $200, or $600 for a larger family, you are getting a product that has had considerably less man hours put into it, probably, than the average application that costs as much.

What are they getting for that? Is all of Photoshop, or Microsoft Office, worth four smallish font families? What's the difference in man hours spent on a few fonts versus MS Office? And what are you getting when you re-buy OT STD fonts? Is it really worth the money?

fonts are rapidly becoming ... like “little applications”.

When they do, and once they have some kind of intelligent interface with the user, maybe users will become more interested in fonts. Maybe that will happen in ten, twenty, forty years.

But -- and I don't intend to pursue this tired point much further -- I know of only one way to engage a user interactively and intelligently with fonts, and that's with LightningDraw style sliders -- right up front, always available -- for varying the appearance of a font. And that means MM or Variations. Until you get the user at that point of simple, fun, engagement, which no application other than LightningDraw ever did, the font is merely a more or less tiresome utility.

The fact that I am one of the few voices in the wilderness willing to criticize Adobe type is an illustration just how unimportant type is in the rest of the world. Who, besides me, ranted for ten years that the opticals in Adobe MMs were badly designed and pleased no-one? Finally, in the OT opticals, we got a bit of grudging redesign. But that came too late to save MMs. Were type really important to users, there would be hundreds of thousands of complaints, not dozens. Hundreds of thousands of complaints: that's, paradoxically, success!

The goal we ought to have is to get fonts to be a product that everybody cares about, not just a few. That means treating them like a competitive software product and productizing them accordingly.

Oh I give up! How many times can one say the same thing?

Christopher wrote: I can’t recall that Adobe has ever replaced a customer’s font for free due to a bug. I’m not sure it’s never happened, but I don’t know of any such case off the top of my head.

It has happened occasionally. There were a couple of cases in the original library conversion to OpenType where we had some issues with usWeightClass settings, which caused significant problems in some Windows applications. I led the push to get that out proactively, and what happened was that every customer who had purchased any of those fonts got emailed and offered the fixed fonts for free. Luckily, it wasn't a whole lot of people.

Regards,

T

Well, Thomas and Chris, who earlier wrote

Thomas talked about the library revision and its bug fixes at TypoTechnica in April 2007

these could be firsts, and mark an extraordinary shift in the wind. I don't want to take away from that.

And let's be fair to the type team: making Adobe's hinting tools available was a very generous piece of philanthropy that even I can't find any conspiratorial loopholes in. (Well, it helps to ensure the continued vitality of PostScript.) Nor is it a one-off, but part of a concerted program of trying to provide as much gratis (and great) type technology as possible. Another public gift of similar magnitude was recently planned, only to be frustrated by a paranoid third party.

Well, well, looking ahead to the future, and to more successful productization of type, all I can say is what I've said before: set up an old Mac with System 7.6-8.1 with the GX extension. Run LightningDrawGX. See how easy, and how much fun, type can be.

Think I'm crazy? The Illustrator team didn't. LightningDrawGX fueled many of the best new features in Illustrator 7, 8 and 9. That's an impressive legacy for a program that was only developed for 18 months.

I'm not saying it's the only way. I'm just saying it's the only way I've seen to date.

Actually, I've frequently discussed bugs in public. Mostly in the Adobe user forums. Somebody speaks up about something they see as a problem, and I comment. That's how we improved our polytonic Greek support last year and got a new beta tester, for example.

Also, many years ago we set things up so that the online bug forms sent in got seen and reviewed by multiple folks, including me. Almost all of them that aren't empty or sent to the wrong team due to user mistake get a personal response, as well, usually from me or Ernie. Again, a minority turn out to be actual bugs, but sometimes they are.

Far and away the most frequent issue we see and hear about is people thinking there's something wrong with the fonts because in "normal" Windows apps not all their fonts show up, due to style linking. That resulted in adding info about style linking to our OpenType User Guide, and to our OpenType readme file, plus revising an existing Adobe KnoledgeBase article. Additionally, we're working on font-specific readmes that will contain family-specific info. The old PDF readmes we had for our Pro fonts also had this info. Plus I've blogged about style linking. Maybe we should also have an article on the main Adobe.com web site... the main problem here is that people just don't read, but we try.

the main problem here is that people just don’t read, but we try.

They don't read, and never will -- that's what we've all been trying to achieve through computer interface design. But there is a solution that will have to come sooner or later, and it should be being architected now: teach fonts how to talk.

Fonts must, in the future, be able to interface with users, if the font industry wants any large success. So there has to be an executable component. What everybody asks is, how on earth does the user interface with that component?

Two ways:

1. Every font name you see in the OS has a down arrow to its right which presents a menu.

2. Right clicking on the font name presents a menu item.

Or, most simply, clicking on a font name, anytime, presents a menu, and doesn't act, merely, as a mechanism for selecting that font.

OK. What's the font going to say to you?

Well, at the simplest level, the font could say hello, I'm Gabbo. I'm a standard OT font and I speak Western European. GPS tells me you're in Roumania. Nice place. Would you like me to tell you what I know about the T-cedilla? OK, and by the way, click here, and for just 1 euro, I'll install the Roumanian component of this font. Hey, you could have Polish and Hungarian for just a couple more euros.

Come to think of it, that's not a bad future model for modular, extensible, software. To participate, you have to set up e.g. a Paypal account. All software comes with modules, in continuous development, that you can add on. A single button press adds the module and debits your account.

This could work for all software, not just fonts. Because people love to shop, especially, and repetitively, for small amounts (hence the phenomenal success of iTunes), it's just about the only constructive, positive, attractive way to counter the idea, hard-wired into all persons under the age of 30, that software is free.

But it has to be made easy! Easier, even, then going to the iTunes store. Just that single button press which gives you a simple choice:

Click me, and I'm yours, for something you can easily afford.

Clarification: when I propose this modular model, it should be clear that the modularity will probably be more virtual than real. That is to say, if the customer click-buys the Roumanian component of Gabbo, what the OS will install is a completely new binary file, not Gabbo WE binary + Gabbo Roumanian binary with the expectation they would somehow work together sans bugs. Similarly, if a customer for 'Basic Word Processor' wants to add some worthwhile feature like multi-level footnoting, I would assume that an entirely new build of 'Basic Word Processor' would be sent to that user. Needless to say, we want to see these virtual modules automatically downloaded and installed. The customer should not have to do anything other than click to buy.

In an era when even clicking on a webpage to go shopping takes up more time than we are willing to give, the attraction of an ultra-simple affordable one-click purchase strategy may prove unexpectedly effective.

Bill,

you're wildly changing subjects as this suits you. This won't help.

1. Adobe last fixed pretty much _all_ of their OpenType fonts in 2005 or 2006 — introducing many fixes to problems that have been reported since the initial release of some of their OpenType fonts in 1999 or 2000. They fixed many of the problems I reported, including some crucial glyph design problems.

2. Once a font is released, many vendors consider it problematic to change the font's metrics (and kerning) because this would cause document reflow depending on which version of the font the user uses. For this reason, for example, after Verdana was mistakenly released with no kerning in the Italic, Bold and Bold Italic styles (kerning was only in the Regular style), it was never fixed.

3. My own opinion is that kerning is one of the most ill-specified elements of OpenType, mostly because it exists in two places: the "kern" table and the "GPOS" table. There are fonts out there that have a "GPOS" table with no "kern" feature while they have kerning in the old "kern" table (Times New Roman, Windows), there are such that have a "kern" feature in the "GPOS" table but not "kern" table (Calibri, Minion Pro), there are fonts that have kerning both in the "kern" feature of the "GPOS" table and in the "kern" table. And, the behavior for kerning is defined differently for .ttf and .otf. And, there are numerous ways to express the same kerning information in a GPOS table (using multiple lookups, multiple subtables, classes or pairs, extension lookup types etc.). I think the "founding fathers" of OpenType screwed up here. Well, the "founding fathers" of TrueType and Type 1 screwed up elsewhere, that's life.

This is all quite complicated for the implementers of OpenType Layout engines. I personally know three major OpenType Layout engines in which the developers tried to implement the behaviors exactly as specified in the spec, but made mistakes because they either did not implement the unspecified "common knowledge" that is shared between Adobe and Microsoft, and is only sometimes described unofficially on some lists.

But the problem as I see it is not that Adobe or Microsoft are trying to sabotage their competition. After all, if it were so, at least OpenType fonts made by Adobe or by Microsoft would work equally well in Adobe and Microsoft applications — but that is not true. The German smallcap ß specified in Microsoft OpenType fonts (Calibri etc.) does not work in Adobe applications, and kerning in those fonts does not in Word, and kerning specified in Adobe OpenType fonts such as Arno Pro does not work properly in Microsoft applications.

But this is not due to ill-made fonts. Most OpenType fonts are made with FontLab Studio plus Adobe FDK plus Microsoft VOLT, and there are THREE people involved in developing the key elements of those applications altogether, so it's easy to make sure crucial bugs get identified and fixed.

But implementing the layout engines — it's not trivial. Not only you need to follow the published specs and the unpublished notes, you need to cooperate with UI designers and convince managers and marketing people that certain aspects of text layout are important, so you will be given a chance to implement it at all. Plus, text layout code has been sitting in operating systems and applications for years, and it's not an easy decision to just "revamp it all".

The one instance I'm actually really angry about is that Microsoft blew the chance to add some proper OpenType Layout features for non-complex scripts when they were working on Office 2007. They changed the file format, for the first time since 1997, so they could have used this chance to look away on the reflow issue. They changed the UI, quite drastically, so they could have used this chance to do away with some ancient font-related options, ancient four-styles-linking, the necessity to explicitly activate kerning and other absurdities of Word, and could have given Word a text engine worthy of the 21st century, with an UI to activate the most important OpenType Layout feature.

And they ended up shipping an app in which for the new default fonts they bundled (Calibri, Constantia, Cambria etc.) there is no way _at all_ to get kerning working, not even if you turn it off, which is actually a step back compared to what was the case with Word 95, Times New Roman and Arial.

I do understand things like release schedules, manpower, deadlines etc. — and yet still, I am angry about that one.

But this issue aside, I think both Microsoft and Adobe have done wonderful job in the last ten years when it comes to original type design, multilingual support, typographic quality, font technology. Hell, imagine that 6-7 years ago, one still had to explain to people that they have to use two different fonts if they wanted to typeset Italian and Hungarian.

Yes, there are flaws, and the industry makes mistakes all the time. Nobody's perfect. I have learned that in the last ten years that are passing since my first ATypI conference. And I have learned to accept that. I do agree with you, Bill, however, that people should acknowledge the actual problems and shortcomings, and "cut the crap".

When Maxim Zhukov was teaching a Cyrillic type design weekend in New York last year, he quoted his former boss at the United Nations who would occasionally ask him the question: "Maxim, is this a screw-up or is this a position?" I very much wish that people would sincerely try to answer this question when it comes to discussion like this.

However, save your ammunition for problems of real magnitude, Bill. And don't shoot at people who are actually on the same team that you are, even if you don't really know it.

Best,
Adam

Adam, this is all brilliantly expressed and anyone who reads it will be grateful for the clear exposition of your viewpoint. I had hijacked the thread for what I thought were some really good new ideas because I thought it was dead, and I am glad you have revived the original topic. But . . . when you write,

I personally know three major OpenType Layout engines in which the developers tried to implement the behaviors exactly as specified in the spec, but made mistakes because they either did not implement the unspecified “common knowledge” that is shared between Adobe and Microsoft, and is only sometimes described unofficially on some lists.

You are saying OT is a broken product that nobody can fix.

Now look at it from my point of view: I saw it all working on the Mac in 1995 in 4MB of memory space. Code was available for both Apple and Windows. Yet nobody wanted it! Neither Microsoft nor Apple. And certainly not customers. Now 13 years later you're telling me to be patient.

I say maybe it's time to start fresh?

Years ago the late Carla Ow-Chu, when she was still at Apple, said, re OT, 'they can't do it without us.'

I said, come on, it'll take them a year or two at most to iron out all the quirks.

Oh no, said. And she was right.

And Adam, I don't think these people are really on my side. I'm on the side of products that work.

Or let me put it another way: here's a situation where a technology vastly affects an enormous amount of application and typeface developers. The technology is spec'd by Adobe and Microsoft. There are two principal problems. One you have identified: they don't share vital information with other app and typeface developers. Two, you were too diplomatic to identify, they don't get on too well with one another.

These two problems are capable of solution. But when? Why not start today?

One thing is clear: no ordinary pressure will work. In eleven years, no ordinary pressure has worked. Only leadership at the top -- and maybe including Apple this time -- can work. Or passive resistance from users. Or else users will respond with sheer indifference.

I just read this entire thread and was amazed at the depth and quality of information posted in it. It's all quite valuable in terms of education and keeping up on the never-ending development/improvement of typography across the aisles, so to speak.

Sadly, as I left the thread and tried to review in my mind what I'd learned, I realized I couldn't steer my mind away from the same unfortunate comment:

"So please, next time do your homework first and stop wasting my time with nonsense."

First and foremost, what might be a "waste of time" to some might be interesting, educational commentary to others regardless of its ultimate veracity. I failed to notice anyone being dragged here by the ear and forced to look at or read anything.

Second, as it turns out the so-called "nonsense" was at least substantial enough to subsequently warrant lengthy discussion ... which must make all of that "nonsense" too, I suppose.

And finally, comments like this seem to echo louder - and for much longer - than others. What's that old axiom - you could have a baseball game in the north part of town, a football game in the west, a basketball game in the east and a fight on the southside and most people will go see the fight?

This sort of disenchanting dialogue was in my mind poisonous to what otherwise was healthy, informative debate. I'm sorry to see that.

Hello Christopher,

some of the people involved in this discussion -- in order of appearance: Miguel Sousa, John Hudson, Thomas Phinney, Christopher Slye, Adam Twardoch -- better than most others know what they are talking about. They have produced the technically most advanced fonts available.
The problem is that Bill ascribed intentions to some of them that make for a grand story but are not backed up by facts. (Which Miguel addressed in straightforward way ... Also note that Bill's last interpretation of Adam's post was kindly ignored.) Indeed, interesting things were said in the course of the discussion, like in later posts by John Hudson, Christopher Slye and Adam Twardoch, yet the length of it was not caused by the matter being so substantial but by Bill's insistence that Adobe does suspicious things. Which simply is not the case:

The trivial reality is that OSes' and applications' OpenType support is still far from sufficient but improving. (Way too slow, if you ask me.) This forces type designers and/or font developers to make tough decisions: Do I add this or that advanced feature, at the risk that some applications won't support it?* Or do I create less than perfect fonts, so they at least behave the same even in less advanced apps?
* Including MS apps which makes it an even tougher decision.

Amazingly, this thread is highly informative despite of Bill's interpretations. And XPress 8 has made it obsolete anyway ...  ;-)

Best wishes,
Karsten

Thank you for your thoughtful comments, Karsten. You have taken a complex thread, which presented an enormous amount of conflicting information, and reduced it all to simple terms without missing out on anything essential. This represents superb analytical thinking.

PS: a few conflicting quotes:

JHudson here: In other words, it is just typical GPOS or GSUB lookup data stored in a different location with a pointer. This is bread and butter to any programmer, and supporting it is pretty trivial.

ATwardoch here: But implementing the layout engines — it’s not trivial. Not only you need to follow the published specs and the unpublished notes ....This is all quite complicated for the implementers of OpenType Layout engines. I personally know three major OpenType Layout engines in which the developers tried to implement the behaviors exactly as specified in the spec, but made mistakes because they either did not implement the unspecified “common knowledge” that is shared between Adobe and Microsoft, and is only sometimes described unofficially on some lists.

JHudson here: So rather than saying ’Adobe should build fonts that work with contemporary products’, I would say that contemporary products should support contemporary text layout, especially when that layout is part of an ISO standard [MPEG!] specification.

JHudson here: I comment from the rock face, and I do consider it the proper business of app developers to build roads through behind me to the place where I am blasting.....pretty much every font I ship only works fully in a small number of current apps, and I expect fonts and apps to develop in tandem ....

JHudson on typophile in 2006: It seems to me that the majority of users are interested in having fonts that are relatively simple to use and which work consistently across a wide range of applications, and this ease of use and compatibility is more important, for most users, than having fonts that do lots of clever things but inconsistently and only if the user knows how to operate them properly.

Couldn't have said it better!

>They have produced the technically most advanced fonts available.

Yeah. So advanced they only work in their developers' heads. Karsten, it isn't about what you can get the font to do in your head. It's about what you can get the font to do in the applications for which it's designed. So the Vista C-fonts don't work in MS Word, not even simple kerning . . . . fonts that work in CS2 are broke in CS3 . . . and on and on. As Adam put it, 'Microsoft blew the chance to add some proper OpenType Layout features for non-complex scripts when they were working on Office 2007.'

Until apps support OpenType consistently and uniformly (i.e., until there is uniform cross-platform OS-level support), OpenType font development is more a hobby than a profession.

Tell me Karsten, when did you last have definitive word from Microsoft about when MS apps are going to support its own OT fonts fully?

If the past is any guide (Verdana roman will never have kerning because it would cause reflow problems with legacy documents, according to John), the answer could well be 'never.'

Anyone know different?

Thomas P: I’ll just note that any kerning that works in TextEdit almost certainly works in other Apple apps such as Pages and Keynote.

Oddly enough while kerning appears to work the same, other features actually make TextEdit the more advanced editor for complex Latin text editing (stacked accents, or simply mark positioning). Try putting a tilde on a q in Pages and then try the same thing in TextEdit.

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

Tell me Karsten, when did you last have definitive word from Microsoft about when MS apps are going to support its own OT fonts fully?

Hello Bill, I can only speak for myself and my two fonts: I simply don't care. It is Microsoft's applications, not mine. If they decide to not support OT layout features -- including kern -- then it is their beer, not mine. Moreover, if people are fine without kerning or without typographic layout features, then I am fine with it too. Office is for office work, not for typesetting books.
And we always have the option to submit feature requests or bug reports.  ;-)

Bill, re. the 'conflicting quotes':

I'd need to go and find the context for the 2006 quote, but I think the viewpoint as stated can be reconciled with what I wrote in this thread by an automotive metaphor. The vast majority of people need cars that are road-worthy, reliable and safe to drive in a variety of weather conditions and on a variety of surfaces. But the development and racing of Formula-1 cars is not unrelated to that need.

Such a nicely expressed thought impels me for once to restraint.