Re: groff 1.25 status (was: [PATCH v2] [eqn]: Use XML character references in MathML output.)

"G. Branden Robinson" <[email protected]> Sat, 4 Jul 2026 06:31:24 -0500

Newsgroups	gmane.comp.printing.groff.general
Message-ID	<20260704113124.pvt2zckhbjt6muou@illithid>

At 2026-06-26T14:54:33-0500, G. Branden Robinson wrote:
> Here's a list of things I'd like to get done for 1.25.
> 
> * Finish refactoring libgroff's `string` class for "fastidiousness"
>   and to seize some performance improvements.  (My working copy has a
>   revision of `string::move()` that appears to buy about a 5%
>   improvement in total groff build time, which if my arithmetic is
>   correct, must be an underestimate, since most of the groff build
>   time is spent running the C++ compiler and linker, not running
>   groff's own programs.  A better measurement would involve scrubbing
>   from the tree the documents groff generates, then rebuilding them.
>   I have a script to do this many times to get a "good statistical
>   sample", but haven't run it on this change yet.)

I've completed this work for now.

The 5% improvement unfortunately proved to be a spurious artifact of
buffering or caching on my system.  The most I can say is that
performance seems unchanged for large workloads like the following.

$ cat ATTIC/64bashes.bash
#!/bin/bash

: ${groff:=./build/test-groff}

set -e

m=$(man -w bash) && \
        for n in $(seq 64)
        do
                cat "$m" >> 64bashes.man
        done

time -p "$groff" -m an -T utf8 -Z 64bashes.man > 64bashes.grout
rm -f 64bashes.man 64bashes.grout

> * Fix Savannah #68345: [troff] stop throwing font deprecation warnings
>   on font names the user remaps to valid names.
> 
> * Fix Savannah #68309: [grohtml] want monospaced fonts back in man
>   pages, gone in 1.22.4.
> 
>   I'm not _positive_ I can fix that one cleanly and correctly, but I
>   want to try.
> 
> * Fix Savannah #68216: [PATCH] tmac/es.tmac: update/correct Spanish
>   string translations.
> 
> * Fix Savannah #66987: [PATCH] [troff] refactor to kill off
>   `ESCAPE_NEWLINE`.

This one's done and pushed.

> * Sort out the eqn(1) MathML generation problem you reported.
> 
> * [Fix Savannah #68407:] Incorporate Stefan Möding's pic(1)ture, as
>   noted above.

This one's done and pushed.  It required significant effort in the build
system; "doc/doc.am" remains a complex beast, but I hope that it is a
more comprehensible monster than it used to be.

There are also some cleanups I'd like to do to "doc/ms.ms.in" and
"doc/Groff-PDF-Features.ms".  That's Savannah #68354.

Lay your wagers now regarding whether I manage to irrevocably commit
transposition errors in ticket numbers 68345 and 68354.

Thoughts?

Regards,
Branden

signature.asc (application/pgp-signature, 833 B)