Re: [groff] 15/39: [grog]: Drop relic code and comments.

[G. Branden Robinson]

[snipping a lot of mutual agreement]

At 2022-11-01T00:08:06-0500, Dave Kemper wrote:
> I didn't see a one-line "require" as bending over backwards,
> but since you pointed out that the version it specified was little
> better than a wild guess, I agree it's inappropriate to retain it.

Well, _maybe_ it wasn't a wild guess.  It's _possible_ it resulted from
a careful consideration of Perl language features and version history at
the time it was put in place.  The problem is that there's no _evidence_
of this being the case, and at the risk of gratuitously dunking on Bernd
again, it would surprise me if it were.

> I can't answer whether it's worth the trouble, but Perl is happy to
> let you wrap a "require" line in an "if" block (or, more perl
> idiomatically: "require v5.6 if cond;").  I don't know the
> build/install system well enough to know what the condition would be.
> Any shell environment variable is available to perl via the ENV{}
> hash, so if one of those has the info you need, the implementation is
> fairly trivial.

I'm glad to hear it's easy, and I have no fundamental objection to doing
this.  I am a little wary of changing behavior between development and
production builds _because_ they're production builds (how many people
really "#define NDEBUG"?), but I could be convinced, if someone is
willing to take our Perl scripts and find the oldest Perl version under
which each one operates correctly--we could then update our GROFF_PERL
Autoconf macro to require the maximum of these.

But at present, I'm not aware of any Perl script we ship that doesn't
work as well as it always has.  In some cases, that's not saying
much--for instance, I've _never_ been able to get glilypond to work, and
I don't know if that's my fault or the script's.  Both the man page and
the code are written such that I rapidly become frustrated with the
author's obsession with trivialities and extensive discourse on things
that aren't the point[1].  I've never yet set aside the time to savage
them as I did grog(1)--which at least had the virtue of working in
_some_ cases so that I could tell if I regressed it.  There have always
seemed to be more important components of groff to work on than
glilypond.  I regret this, because as an amateur musician I'd like to
learn LilyPond, and embedding its output in groff documents appeals to
me far more than messing around with TeX.

Regards,
Branden

[1] Remind anyone of my emails?

signature.asc
Description: PGP signature