News Analysis: First of two parts: PDFzone gets an inside look at how Adobe tests new versions of Acrobat and Reader and what happens when an incompatibility issue comes up.Anyone in charge of a company's PDFs—archives of vital business documents, magazine issues, or profit-making content sold to subscribers—is likely at some point to wake up in the middle of the night after a nightmare that goes something like this: After years of smooth upgrades, Adobe releases a new version of Acrobat and Reader, and the files suddenly don't open correctly.
Your company's PDFs are still there, but they're maimed. Characters don't render right, error messages pop up. A line of angry people—armed with staplers, rapier-sharp letter openers, and vessels of scalding hot coffee—forms at the mouth of your cubicle. And they want answers.
For one of PDFzone's readers (unnamed here because he works for a large company whose strict corporate policy is never to comment on other companies' products), there was no waking up.
The nightmare became his day job when Acrobat and Reader 7 wouldn't properly open some of the 6,500 PDFs—niche publications filled with expensive technical content—that his company sold to customers around the world.
Specifically, users of documents this company created in Distiller 3.0.2 in the late 1990s got a "Cannot extract embedded font" error, and while the documents opened in Acrobat and Reader 7, crucial math characters didn't display—rendering the PDFs useless to customers.
This brings us to the eternal debate that rages on PDF-related message boards: Should Adobe make the PDF spec a totally open standard developed jointly by Adobe and an industry-wide standards committee? What about Reader and Acrobat?
One side of the debate says that since public institutions such as governments are legislating the use of PDF as their exclusive format for electronic documents, shouldn't these technologies be public property?
As it stands now, Adobe can do what it wants—up to and including making future PDF specs completely private and proprietary—and hold us all hostage until we pay up.
The other side says no way should PDF and Acrobat be entrusted to Joe Everyprogrammer. Basement developers will kill PDF faster than you can say "Apple Lisa."
A commercial company like Adobe, with its army of software engineers, will make a far better product than any consortium of volunteers devoting their spare time to concocting Acrobat and PDF products.
Big companies like banks and insurance companies need an Adobe to dictate PDF security and stand accountable for the stability of Acrobat and PDF.
In a sense, both worlds exist now: Freeware, shareware, and commercial wares all write and render PDF files of varying quality.
Adobe makes the flagship products, which brings us back to our story. The angry calls and emails rained in and our protagonist faced some grim choices:
Tell his customers to downgrade to version 6 of Reader and Acrobat (for some sites, that involved hundreds of PCs)
Run the offending PDFs through Acrobat 6 Pro, which paperlessly "scans and OCRs" the files a la PDF Capture, which, in his opinion, significantly degrades the content
Recreate the files from scratch
Seek a third-party solution, cross your fingers, and hope it works
The first three choices were either logistically unworkable or cost-prohibitive. The fourth choice would have worked perfectly, but for the bugs in the third-party filter that fixes the PDFs.
Next Page: Support problems.
So he was left with contacting Adobe and hoping for the best.
While at first blush the issue might seem as simple as "fix it or don't" for Adobe, it's not.
The problem with supporting PDF—which is a public standard in the sense that anyone can get hold of the spec—is that you have to support everybody's PDF.
No matter if it's created by quite stable competing commercial products such as Global Graphics' Jaws PDF; questionably stable tools with names like Joe's Basement FreePDFWriter; or in this case, the crusty old classic Adobe Acrobat Distiller 3.0.2.
Adobe also finds itself supporting everything that gets thrown into a PDF, such as error-ridden fonts. (No one's yet figured out where this story's font in question came from, whether it was an Adobe font, one from another type foundry, or a custom job created in-house. It's just there in the PDFs, embedded.)
Adobe knew about the problem, or at least Acrobat and Reader 7's reaction to it, says David Stromfeld, one of several Acrobat product managers who also is familiar with this particular Reader issue, even though Reader's been moved to a different team since the Adobe-Macromedia merger.
In versions 6 and previous, Adobe allowed a certain tolerance for font errors that engineers tightened up in version 7.
"In the course of testing [PDF files] for Acrobat 7, we learned of a particular attribute that could appear in fonts that basically cause instability in the application—it could actually crash Acrobat—and cause instability to the user's system," Stromfeld says.
"This was a particular attribute in a font—an error in font creation—and because it could affect the user's system, we changed the way Acrobat handled that particular type of font that had that attribute."
Through 2005, after the release of Acrobat and Reader 7, Adobe received a few reports about the error dialog, but not so many that it felt compelled to reverse course and allow those PDFs to open as they had before. However, this story's PDF guy, working for a well-known company, and his potentially 6,500 problem files, proved to Adobe that the change probably affected more than just a few users.
Adobe probably could devote all of its engineering bandwidth to chasing down potential problems like this, no matter whose font or PDF writer caused them.
Of course, that would leave no time for developing future versions of Acrobat. So where did that leave our PDF guy, his files, and the customers who depended on them?
Read part two of this two-part article series.
