Detect rare situation where the footnote caller gets pushed to next/final page

[markpenny]

A PDF we were discussing had an interesting end-of-book situation where the footnote caller got pushed to the final page of the book, while the footnote stayed on the previous pg. Given that this is unacceptable, I feel that we should at least alert the user to the fact that this has happened so that they can take measures to solve the chicken-and-egg problem some other way, for example by shrinking previous pages to make it all work as it needs to.

Along the same lines, it might be possible/appropriate to warn the user of pages that have large amounts of unintentional white-space on them.

[davidg-sil]

An archive is probably necessary so that we can identify how this happened. It certainly shouldn't! Possibly it was something to do with it being the final page. A potential solution:

• TeX should maintain a counter is kept which uniquely identifies each footnote, endnote, etc
• Code to write the page number and that unique identifier to a check-file (possibly parlocs) for both the caller and the note text. (possibly also the end-of-note, to detect notes that have been split)
• Code to check that the caller and callee are on the same page, and moan if they are not, or indeed, moan if a footnote has not appeared in the output at all.

Thoughts/problems with this:

• In a study bible, there are a lot of notes / xrefs per page.
• It will be necessary to keep a list of callers and 'cross them off' as they are met - this could end up as O(N^2) or worse if we get it wrong.
• End-notes will need to either be excluded or (if we want to ensure that they are not lost), the list carried over from page to page.
• If the handling of this is done without in purely in TeX, the memory or processing requirements could become very significant unless we can deallocate things properly.
• Is there any benefit to processing them in TeX? Is there ever a need to format a footnote differently on odd versus even pages, for instance? If not, would it be better to process the file via python?

[davidg-sil]

The unique ID could become some kind of PDF anchor, giving clickable send/return feature. Is that a good thing?

[markpenny]

All good questions! Yes, I am almost certain that this was as a result of being the last page of a long book, and something different happening there compared to the rest of the publication. The PDF is still in the Archive dropbox if it helps: "Laarim NT+ LOH 4810 Body_LowRes.pdf", pg 1041 is the page in question, footnote i at PHP 4:22.

If TeX generated the file with the data, it would be relatively straightforward for Python to parse and analyze it. That's probably a much neater solution that trying to implement it in TeX.

[davidg-sil]

The TeX side of things is now partly functional - It cannot work with old-style paragraphed notes without destroying the output. New-style paragraphed notes are OK. Python or TeX code to check that the results are sensible is the next step.