two research challenges for the curious

[G. Branden Robinson]

I've stumbled across two little surprises lately and was wondering if
anyone wanted to look into them.  You don't need to be running
bleeding-edge Git; groff 1.22.4 will do.

(1)

Try any of the following commands:

zcat $(man -w groff_mdoc groff_me) | groff -mandoc -t -Tutf8 | less -R
zcat $(man -w groff_mdoc xtotroff) | groff -mandoc -t -Tutf8 | less -R
zcat $(man -w groff_mdoc ls) | groff -mandoc -t -Tutf8 | less -R

Then scroll to the beginning of the SECOND page rendered.

You'll notice that the last text line of groff_mdoc(7) has been
startlingly emitted at the beginning of the next page, whatever that
page might be.

I think this may have something to do with diversions and output lines
not being full, but I'm not sure.  I still don't feel like I really
understand diversions.

(2)

Review this bug:

https://savannah.gnu.org/bugs/index.php?44708

I'm not positive, but I think this might actually have been another
issue in disguise, one that John Gardner noticed[1] after I started
merrily bracketing our man pages and macro files with compatibility-mode
save and restore pairs if they used any groffisms at all.  As you may
recall, the fix for that was to expose the state compatibility mode is
"really" in, even while .do is temporarily hiding it.  See commit
6a37bb5f00d881e50856dc5906f03e64dcacd1ec [2].

Then check commit 42c866a3fc0552f7f91a4af1ba93a942130e1bd7 [3].

I think the latter might be revertible, in principle even if it's not
practical to do a "git revert" after 5 years.

Anyway, if no one beats me to it, I aim to experiment with that a bit.

Regards,
Branden

[1] https://lists.gnu.org/archive/html/groff/2020-04/msg00027.html
[2] 
https://git.savannah.gnu.org/cgit/groff.git/commit/?id=6a37bb5f00d881e50856dc5906f03e64dcacd1ec
[3] 
https://git.savannah.gnu.org/cgit/groff.git/commit/?id=42c866a3fc0552f7f91a4af1ba93a942130e1bd7

signature.asc
Description: PGP signature