Re: [NonGNU Elpa] New package: ETT

[Eli Zaretskii]

> From: John Task <q01@disroot.org>
> Cc: emacs-devel@gnu.org
> Date: Wed, 03 May 2023 14:14:50 -0300
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Anyway, did you consider extending existing packages with those modern
> > features, instead of making a completely new separate package?
> 
> [...]
> I get your intentions, but all things considered, I think everything
> is better as is.

>From your POV, no doubt.  I'm well familiar with the joy of designing
my own package or feature from the ground up, unburdened by history,
backward compatibility, and all that ballast.  If nothing else, it's a
lot of fun.

But from the project management POV (which is the hat I'm wearing
now), it is a problem if, instead of developing and extending existing
packages, we keep introducing new and different ones.  It makes Emacs
as a whole harder for users to learn and follow for years to come:
they can never be sure the packages they use and are familiar with
will not be abandoned and replaced by others which they will need to
learn from scratch.  It also presents hard maintenance dilemmas for
which there are no good answers: should we have multiple packages with
partially overlapping functionalities, or should we abandon the old
one and embrace the new one(s)?

This is why extending an existing package, although it is harder and
requires more efforts from the developers, is always preferred here to
additions of new and incompatible ones.  E.g., timeclock could have a
"time-tracking" feature or mode added to it, where it would target the
usage patterns you had in mind, rather than the original purpose of
timeclock (which still has its merit, btw, and many smartphone apps
which do just that are a living proof of that).  And, of course,
statistics and graphs can be added to any existing feature, it doesn't
IMO justify a new package written from the ground up.

Of course, I'm too old to naïvely believe that any of the above will
convince anyone to scratch a different itch.