Bottom margin vs. descenders question

[Shreevatsa R]

On Sun, 5 Jul 2020 at 09:54, Doug McKenna <doug at mathemaesthetics.com> wrote:

> I have concluded with if all one has is a PDF and a nominal rule that
> there should be a 1" bottom margin on the page with no page numbers, it's
> not possible to report a margin breach robustly without false positives,
> due to descenders generally being allowed to go below the margin's bounds,
> which makes optical sense.  It works for unwanted page numbers, though,
> which is worth complaining about in a paper submission.  The proceedings
> editors will still have to do some visual culling of occasional false
> positives.
>
> Enclosed is a screenshot of what I have tried to output in my program.
> The red lines have been superimposed onto the PDF to show where the margin
> bounds are.  This seems a perfect example of what has just been discussed:
> a math formula with large delimiters on the bottom line of the page that
> descend quite a bit into the margin area in order to maintain the baseline
> of the bottom line on every page at the same vertical position.  It is
> typeset properly, even though it breaches the margin.  I believe this was
> typeset with TeX/LaTeX.
>

It seems what you could do is to (try to) detect the lowermost *baseline*
on the page (via some combination of information from the lowermost
y-position with sufficient proportion of black/ink, or trying to detect the
inter-baseline skip and using that somehow), and then report a violation if
this lowermost *baseline* intrudes into the 1-inch margin. (I believe some
PDF-related programs like k2pdfopt do have heuristics to detect baselines
like this.) I don't know how reliably this could be done, so I guess it
introduces some risk of false negatives as well, but depending on the
current false positive rate (caused by descenders), this may or may not be
worth it.
On second thought, my guess is that this would not be worth it -- too much
work for too uncertain an outcome; better to get the users to understand
baselines, and accept the false positives from the simpler and
well-understood process.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/texhax/attachments/20200706/b099f757/attachment.html>