Font problems after 10.6.7

Very interesting, John. I downloaded FontXChange and converted the entire set of Myriad Pro. It let me know that I was converting a font that was already in the same format I was going to (OT PS).

Mostly, I wanted to check if it was really creating an .otf font that was actually TrueType, but it's not. They are PostScript.

Can't check it right now, but when I get a chance, I'll try the converted fonts in 10.6.7 and see what happens. Even if it works, the issue really is that we shouldn't have to.

etresoft wrote:
They did not test with OpenType Postscript fonts in Acrobat. They probably did test lots of other things with Acrobat, but didn't use Postscript fonts.

Do you have any certain knowledge about that or is it just a guess? People seem to be largely ignoring a not so small elephant in the room: the reports of font problems are not consistent. Some are saying all OpenType Postscript fonts are affected, others only some of them. This suggests there may be another, as yet undiscovered factor involved, one that isn't the same for all systems.

It seems highly unlikely to me that Apple or its beta testers never tried any of these fonts with Acrobat, or for that matter that Adobe (which almost certainly does its own tests with developer prerelease versions of OS X updates) didn't either. It seems much more plausible that they just didn't test with systems affected by the unknown factor(s), whatever it or they might be.

That in turn makes me wonder how many users actually are affected. There is a tacit assumption being made that everyone either is or would be if they tried an appropriate test, like attempting to open a PDF file created by the 10.6.7 Quartz PDF generator with Reader, but as has been pointed out that should be so common that it would apply to a very large number of users. It is hard to gauge such things but it seems like there may not be enough reports to justify accepting this assumption unconditionally, especially together with the inconsistencies in the available reports.

Regardless, apparently Apple & Adobe are cooperating in looking for the cause & the solution. That may or may not result in a change in 10.6.7's PDF generator, something else that it may be unwise to make too many assumptions about.

canonballs wrote:
When 10.6.7-produced pdfs cannot be opened on 10.6.6 it is difficult to argue that Adobe alone is to blame.

Except that this doesn't seem to be a common problem. In fact, there are so many different problems being reported, some conflicting with others, that one guess is just about as good as another about who is to blame for them.

Besides, for things like this sometimes nobody in particular is to blame. Sometimes standards just aren't as well defined as one might expect, leaving grey areas that are subject to "best guess" interpretations. If everybody's guesses are about the same, it is easy to incorporate them into a revision of the standard & most users aren't even aware there was a potential problem. But if they aren't, as long as they don't actually violate anything in the existing standard, none of them are "wrong" in any meaningful sense & there is no one to blame.

rdhw wrote:
Please show your evidence for these assertions. To which Adobe "bug" are you referring?

My evidence is that PDF implementations that are known to be independent of Adobe can all display these documents correctly.

The "bug" to which I am referring is the inability to open an apparently valid PDF file. Out of 4 independent PDF viewing programs I tried, only one crashed - Adobe. Unfortunately, they have at least 95% of the market. That is why Apple has to take responsibility for it.

Please explain why products with no Adobe code in them at all now fail under 10.6.7 with OpenType PostScript fonts, when they worked under all previous versions of Mac OS. (For instance, Sibelius is not able to display scores correctly on the display screen if elements of the score are taken from OpenType PostScript fonts).

You don't know that Sibelius doesn't use Adobe code. Adobe has licensed this code for years so that other products can easily use Postscript and PDF. It seems reasonable that software costing hundreds of dollars (such as Microsoft and Avid products) is more likely to be able to license Adobe code than small companies or open-source products. The ironic thing is that said small companies and open-source products have no problem reading these 10.6.7 PDF files.

R C-R wrote:

canonballs wrote:
When 10.6.7-produced pdfs cannot be opened on 10.6.6 it is difficult to argue that Adobe alone is to blame.

That in turn makes me wonder how many users actually are affected. There is a tacit assumption being made that everyone either is or would be if they tried an appropriate test, like attempting to open a PDF file created by the 10.6.7 Quartz PDF generator with Reader, but as has been pointed out that should be so common that it would apply to a very large number of users. It is hard to gauge such things but it seems like there may not be enough reports to justify accepting this assumption unconditionally, especially together with the inconsistencies in the available reports.

We have read reports by rcberwick and schmidhauser. The first tells us that this problem is both widespread and reproducible. The second was accompanied by a 10.6.7-modified file that neither I nor anybody on this thread who tried it can open on 10.6.6 with either Preview or Adobe Reader. That file contains nothing but bog-standard Computer Modern (PS type 1) fonts that have been around for more than a decade. I have not seen any reports that contradict these findings. So yeah, I tend to believe them.

R C-R wrote:
... this doesn't seem to be a common problem. In fact, there are so many different problems being reported, some conflicting with others, that one guess is just about as good as another about who is to blame for them.

This may be just one aspect of the problem, but for me it is scary enough on its own merit. It's a little bit like finding that your library has been infested by bookworms (not in a metaphorical sense) and threatens to become unusable. As for it not being "common", yes I can imagine users who won't be affected but I can assure you there are lots and lots of users who will. You could probably get a crowd large enough to bring down a government or something if they brought pitchforks.

I wonder how you would feel about the same thing happening to files in .aiff or .jpg format. (I do appreciate that these are simpler than .pdf.)

Besides, for things like this sometimes nobody in particular is to blame. Sometimes standards just aren't as well defined as one might expect, leaving grey areas that are subject to "best guess" interpretations. If everybody's guesses are about the same, it is easy to incorporate them into a revision of the standard & most users aren't even aware there was a potential problem. But if they aren't, as long as they don't actually violate anything in the existing standard, none of them are "wrong" in any meaningful sense & there is no one to blame.

I appreciate this point. A few years ago I was myself impressed by this when looking at EXIF 2.21. Examples above however suggest (I'm not saying beyond reasonable doubt) that in 10.6.7 and 10.6.6 Apple may have followed interpretations of the standard that are inconsistent with one another.

R C-R wrote:

etresoft wrote:
They did not test with OpenType Postscript fonts in Acrobat. They probably did test lots of other things with Acrobat, but didn't use Postscript fonts.

Do you have any certain knowledge about that or is it just a guess?

It is obviously just a guess. I am allowed to make educated guesses.

People seem to be largely ignoring a not so small elephant in the room: the reports of font problems are not consistent.

Yes - on purpose. When I first heard of these reports I labeled them nonsense. To prove my point, I upgrade to 10.6.7 and tried it myself. Imagine my surprise when I was able to reproduce the problem very easily. I created super-simple PDFs in both 10.6.6 and 10.6.7 and analyzed them byte-by-byte to see what the differences were. There were two differences: (what follows is a quote from my Apple bug report)

1) The /Gs1 and /Gs2 parameters to the ExtGState command are swapped in 10.6.7. Adobe Reader for
Windows does not like this, even though the objects these references point to are also swapped. This
change alone is enough for Adobe Reader to refuse to open the file.

2) The embedded font is compressed differently in 10.6.7 than it was in 10.6.6. Even though both data
streams decompress to identical data, Adobe Reader does not like the new version in 10.6.7. At best,
this causes Adobe Reader for Windows to report an error extracting the embedded font. For more
complicated PDF documents, Adobe Reader crashes.

It seems highly unlikely to me that Apple or its beta testers never tried any of these fonts with Acrobat, or for that matter that Adobe (which almost certainly does its own tests with developer prerelease versions of OS X updates) didn't either. It seems much more plausible that they just didn't test with systems affected by the unknown factor(s), whatever it or they might be.

As unlikely as it may seem, it seems to be the truth. The PDF changes seem buried in an obscure security update portion of 10.6.7. They corrected some buffer overflow issues that apparently Adobe code relies on. I was able to reproduce this with a one-line PDF file. My byte-for-byte analysis wasn't hard at all because the resulting PDF files were so small. It does appear that no one tested this with OpenType Postscript fonts. My theory is that Adobe's bloated code users entirely different logic when dealing with Postscript fonts as opposed to Truetype fonts. The Truetype code uses newer logic that will accept the compression scheme that Apple is using. The older, Postscript code hasn't been updated in years and will only support a Postscript font compressed circa 1996. I don't fault Adobe for that - if it isn't broken, don't fix it - and also - if you have 100 million installations, don't fix it.

That in turn makes me wonder how many users actually are affected. There is a tacit assumption being made that everyone either is or would be if they tried an appropriate test, like attempting to open a PDF file created by the 10.6.7 Quartz PDF generator with Reader, but as has been pointed out that should be so common that it would apply to a very large number of users. It is hard to gauge such things but it seems like there may not be enough reports to justify accepting this assumption unconditionally, especially together with the inconsistencies in the available reports.

That is because you only encounter this bug when using OpenType Postscript fonts.

Regardless, apparently Apple & Adobe are cooperating in looking for the cause & the solution.

Do you have any evidence of that? One of the things that seems highly suspicious is that I can find no posting in the Adobe forums by an Adobe employee blaming Apple for this bug. That is a first. They have never been this shy before about pointing the finger at Apple.

Do you have any evidence of that?

This the only known official statement to date where it says Adobe is working with Apple on a solution.

http://kb2.adobe.com/cps/896/cpsid_89637.html

While this particular link refers to Flash, the same stretched OT PS font issue has been reported in Final Cut Studio and other apps. This along of course with the barrage of posts on unusable PDF files.

etresoft wrote:
It is obviously just a guess. I am allowed to make educated guesses.

Sure. But even educated guesses need to account for all the facts to be credible, including those we would prefer to ignore. 😉

The real challenge is to explain why what is reproducible on one system is not necessarily reproducible on another. For instance, if a pair of swapped parameters even in a super-simple PDF is the underlying cause, why is it that some users are reporting that only certain OT PS fonts cause problems, or even that it takes two or more fonts from the same OT PS family to trigger them?

You are assuming that all reports on Apple Discussion constitute "facts". People are wrong and get confused all the time. I think the last time I got confused and started to spit out misinformation was 2 days ago in this very thread or maybe the other one on the same topic.

It is not possible to determine "facts" from Apple Discussions. You have to take your own experience, add a fair amount of logic, and come up with a "best guess". I will use that to provide solutions and workarounds for the problems as I think they actually exist. Hopefully that helps some people. That's the best I can do.

I am having the same issue, using Myriad Pro in Final Cut, Motion seems to handle it fine, but Final Cut needs to be re-booted every once and and a while to get it fixed.

A warning about using FontXchange in trial mode: It makes random changes to some glyphs. There's a notice telling you this when you click on the demo button.

Check your proofs very carefully if you are using the FontXchange demo.

Thanks, Stevie. I did see that and wasn't concerned. I'm only going to use the converted fonts for testing, whenever I get to that. But good info to know so people don't think, "Well, I'll just use it for free instead of paying for it."

Is anyone else seeing the behavior of various fonts changing over time? When I first became aware of these problems, I went to one of my 2 machines that I've upgraded to 10.6.7 and created some pdf's. All of these are created via TextEdit, setting some minimal test text into a font, printing to PDF from the File->Print panel, and then dropping the resulting pdf files into Adobe Reader 8.1.4.

I have a pdf file that was created on Mar 29. It is Minion Pro Regular. When I drop it in to Reader, I get the error message "Cannot extract the embedded font 'BTANIW+MinionPro-Regular'. Some characters may not display or print correctly." The file appears, and all of the characters appear as irregularly spaced bullet points.

I followed the same steps on April 6 and created a pdf file that says "This is a test" in Minion Pro Regular. If I drop this file into Reader, no error messages, and it displays in what to my non-expert eyes appears to be Minion Pro Regular.

Thoroughly perplexed, I changed the text in my file to Minion Pro Bold Cond Italic and plopped out a pdf. I drop this into Reader, error message: "Cannot extract the embedded font 'TBP5GM+MinionPro-BoldCntIt'. Some characters may not display or print correctly." And my file appears as bullet points.

Ok, another day, another test. This time I create a file with the words "Minion Pro Bold Cond Italic" and put it in the font Minion Pro Bold Cond Italic. Drop into Reader, and I get the error message: "Cannot extract the embedded font 'EPURRF+MinionPro-BoldCntIt'. Some characters may not display or print correctly." This time, instead of the bullets, the text prints out, but it is in a san-serif font. (I'm no expert, but I can tell the difference between serif and san-serif.)

Next test was to take my file and type in
Minion Pro Regular
Minion Pro Bold Cond italic
Myriad Pro Bold Cond italic
Myriad Pro Regular
I then put each line in the font that matched the text. Dropped the PDF file into viewer, and get the error: "Cannot extract the embedded font 'CZNQJP+MinionPro-Regular'. Some characters may not display or print correctly." The Myriad Pro text might be correctly displaying in Myriad Pro, but the Minion Pro stuff is in a san-serif type, and the bold isn't as bold as I would expect, and it's not condensed.

Color me perplexed...

Same issue, happened today, both platforms resulted in a "hang" when trying to print Pdf forms from Turbo Tax. Solved this by printing to iphoto. I initially thought it was something wrong with the printer drivers until I read your post.

I work for a small desktop publishing company that contracts out to bigger educational publishers. We are also having this problem. Luckily we only have two computers with 10.6.7

I have spent a long while reading this thread and getting an understanding of what the problem is. Am I correct in summarizing this: The only solution at this time is to downgrade to 10.6.6

Is that correct? I just want to make sure I am following this very long thread 🙂

Thank you to everyone who is posting and spending so many hours to help clarify this issue.