Follow-up Comment #1, bug #68345 (group groff): On Wednesday, 13 May 2026 09:05:44 BST you wrote: > URL: > <https://savannah.gnu.org/bugs/?68345> > > Summary: [troff] stop throwing font deprecation warnings on > font names the user handles by remapping to valid names > Group: GNU roff > Submitter: gbranden > Submitted: Wed 13 May 2026 08:05:39 AM UTC > Category: Core > Severity: 4 - Important > Item Group: Warning/Suspicious behaviour > Status: Confirmed > Privacy: Public > Assigned to: gbranden > Open/Closed: Open > Discussion Lock: Unlocked > Planned Release: None > > > _______________________________________________________ > > Follow-up Comments: > > > ------------------------------------------------------- > Date: Wed 13 May 2026 08:05:39 AM UTC By: G. Branden Robinson <gbranden> > Spawned off of bug #68261. > > There, in comment 9, Deri wrote, in reply to me: >> Whilst it >> [https://cgit.git.savannah.gnu.org/cgit/groff.git/commit/?id=5011cd1d5a >> commit 5011cd1d5a] >> is only a deprecation and works (by switching to CR as it always >> has), I'm not too bothered, it is educating the users away from using a >> pseudo >> font (and the actual warning can be suppressed as you explained). My >> concern was if, at the end of the deprecation, you intended to convert >> the warning to >> an error, since it would mean that users who decide to tell troff which >> font to use instead of the pseudo (i.e. have heeded the warning by >> adding:- > .if t .if \n(.g .ftr C CR > >> To a startup file to point troff to a real font) will be miffed. > > I had no plans to "convert" the warning to an error, but instead to drop the > specialized deprecation code introduced by that commit. Which, however, > would probably amount to the same thing. If a user tried to select font > `C` or `Hx`, and didn't have font description files set up for fonts of > those names for the selected output device, then yeah, they'd get errors, > just as in the following case. >
I thought this would be your action, better to check. :-)
> $ echo '.ft BOGUS' | ./build/test-groff -T ps
> troff:<standard input>:1: error: cannot open font description file 'BOGUS':
> No such file or directory
> troff:<standard input>:1: warning: cannot select font 'BOGUS' [-w font]
>
>
> In fact, what you propose would make GNU _troff_ **more** self-consistent:
>
>
> $ printf '.ftr BOGUS TR\n.ft BOGUS\n' | ./build/test-groff -T ps 2>&1 \
>
> | grep . || echo cool beans -- no diagnostics
>
> cool beans -- no diagnostics
>
>
> Contrast:
>
>
> $ printf '.ftr Hx TR\n.ft Hx\n' | ./build/test-groff -T ps 2>&1 | grep . ||
> echo cool beans -- no diagnostics
> troff:<standard input>:2: warning: font name 'Hx' is deprecated [-w font]
>
>
> So, yeah, I think I had better change this.
>
> Setting Severity to Important because I did not intend for GNU _troff_ to
> screech at the user when they take reasonable steps to make their document
> not select nonexistent fonts...but since we're screeching _before_ applying
> font translations instead of after, that is the outcome.
>
> That means moving the deprecation warning higher in the call stack. I'm a
> little nervous that this will mean having to change function signatures to
> add an argument, or making `is_device_ps_or_pdf` global, but we'll see.
>
>
_______________________________________________________
Reply to this item at:
<https://savannah.gnu.org/bugs/?68345>
_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
signature.asc
Description: PGP signature
