Follow-up Comment #3, bug #66165 (group groff): Okay, I found the problem. I instrumented line 292 of "pdf.tmac" and it showed right up.
$ ./build/test-groff -UZ -T pdf ./ATTIC/woolly-faced-neanderthal.groff | grep '^x X' x X ps:exec [/Dest /pdf:bm1 /View [/FitH 5001 u] /DEST pdfmark x X ps:exec [/Dest /pdf:bm1 /Title (Introduction) /Level 1 /OUT CRO-MAGNON FREAKOUT pdfmark x X ps:exec [/Dest /pdf:bm2 /View [/FitH -31000 u] /DEST pdfmark x X ps:exec [/Dest /pdf:bm2 /Title (My day with a woolly-faced neanderthal) /Level 1 /OUT CRO-MAGNON FREAKOUT pdfmark x X pdf: pdfpic ATTIC/woolly\[u2010]faced\[u2010]neanderthal.pdf \[u2010]L 468000z 605880z And what is that line, once my cro-mag nonsense is removed? $ sed -n '292p' tmac/pdf.tmac . nop \!x X ps:exec [/Dest /\\*[PDFBOOKMARK.NAME] /Title (\\*[pdf:title]) /Level \\n[pdf:bm.lev] /OUT pdfmark Observe the gymnastic achievement of a device extension command via *neither* the `\X` escape sequence *nor* the `device` request, but raw grout stuffed into a "transparent throughput" escape sequence! Why such indirection? [https://git.savannah.gnu.org/cgit/groff.git/commit/?id=4536678ce5713907304ad1695e907f6888c44aa3 It looks for all the world like a quick hack way back in 2012 to get around problems with the "pdfmark" package, or maybe with _troff_ itself.] -. pdfmark /Dest /\\*[PDFBOOKMARK.NAME] /Title (\\*[pdf:cleaned]) /Level \\n[pdf:bm.lev] /OUT +. nop \!x X ps:exec [/Dest /\\*[PDFBOOKMARK.NAME] /Title (\\*[pdf:cleaned]) /Level \\n[pdf:bm.lev] /OUT pdfmark +.\". pdfmark /Dest /\\*[PDFBOOKMARK.NAME] /Title "(\\*[pdf:cleaned])" /Level \\n[pdf:bm.lev] /OUT The retention of that pdfmark macro call as a commented line is suggestive. I'm guessing that either the "pdfmark" macro was trying to do too much (a safe bet in my experience), and/or that using `device` or `\X` directly had undesirable side effects. "Fortunately", GNU _troff_ was inconsistent and the nowhere-documented differences between these disparate means of stuffing bytes into grout was usable to one's advantage here. And thus a quick hack settled in for a 12-year residency without the problems provoking it getting addressed. This hydra-headed beast of inconsistency is what I'm trying to kill in bug #63074. _______________________________________________________ Reply to this item at: <https://savannah.gnu.org/bugs/?66165> _______________________________________________ Message sent via Savannah https://savannah.gnu.org/
signature.asc
Description: PGP signature
