Re: groff: grops and grodvi crash on invalid input

[John Gardner]

>
> He declared a groff-compatible ps device resolution of 72000 but didn't
> scale his arguments.


That's assuming he didn't patch devps/DESC locally with an arbitrary `
sizescale` or `unitwidth` parameter (which may have been an attempt to make
coordinates easier to visualise). But your point still stands.


That's documented in groff_out(5) as against the rules.


IMHO, it would be more intuitive to make this a no-op, rather than an error
(a warning at *most*). Users may wish to "prepostprocess" the formatted
output to control page order, extract or inject existing pages from an
earlier formatting run (etc)… things that elevate the risk of an errant
motion command appearing before the first p command.

Roff.js, being designed to accommodate a range of troff implementations and
perform unattended document formatting, is deliberately tolerant of input
it doesn't expect (to a fault: I actually need to make it a little stricter
with error reporting... I'll get back to that in 2021 probably…)


Now this part is really weird.  I don't know where those came from.


Probably Gmail. Google probably converts the line-endings of *.txt
attachments to make life less shit for Windows users, since less
tech-literate people won't understand why Notepad.exe is displaying
everything on "one huge line".