I've read many complaints about ID's limited footnote capability on this forum and elsewhere, but I recently had a problem I don't recall seeing before: export to PDF corrupted footnote numbers, in no pattern I could see.
My work is scholarly material with lots of footnotes, and I've been using ID seriously since version 2 (when footnotes still had to be done by hand à la PageMaker). Final output is generally print, though some jobs also want PDFs for electronic distribution. I mostly use tried-and-true IDCS4, though I have later versions installed on a test-bed. I'm fully updated, under 64-bit Win7 (Intel i7-2600 with 16 GB RAM). I like to proof in PDF, and have generated untold thousands. This week, for the first time ever, ID's export to PDF screwed up footnote numbering on some 80 pp. in a 500-pp. book. Has anyone else seen this?
In the bad PDF, the footnote numbers in the text (re-started from 1 by perhaps 18 ID "sections") are fine, but on about 15% of the pages (perhaps 30% of the pages with footnotes) the numbers in the notes had increased their values by 2 or 3. The corruption was not easy to spot, and once a few errors were noticed I needed to assess the damage: I generated a new PDF from a duplicate file, with the same Press settings, and used Acrobat 9 Pro's text comparison against the bad file to find all the differences.
Both the original ID file and the duplicate look fine on screen, suggesting the problem occurred during this one export. The job is a tricky one, a single ID file to accommodate some 2,100 cross-references. It began as a Unicode text file I imported into IDCS4, where I used GREP to convert ASCII tagging to InDesign formatting (italics, etc.). The file had shown some instability, occasionally reporting a dozen cross-references spontaneously going bad. Passing the file through IDML did not cure this (I assume INX would strip the cross-refs). However, stability returned when I replaced the one piece of art, an EPS vector, with an AI conversion.
I'd like to think this was just a quirk of that one instance of exporting, perhaps because a leak somewhere reduced memory available to ID. Or do I need to run an Acrobat comparison before I send anything out?