Being fine-grained has never been a goal.
That is the problem.
Being more granular is the _solution_.
Nobody asks "what fonts should I use".
Yes they most certainly do. In fact, after "what's your favorite font", it's the most common question you get from desktop Linux / FOSS users once they hear you're a 'font person'.
Maybe the question should be
rephrased to "what set of fonts should I have on the system so I feel
the least urge to immediately install more other fonts?".
Here again is point: the user does not feel this sort of urge in a vacuum. Instead, they feel it once they start working on some sort of document (be it code in an editor, slides or papers in an office suite, or decorative matter in an image editor).
It's clear to
all of us that this cannot cover the most exotic use cases but merely a
common denominator for the average user.
Perhaps it's up to who is considered "us" in this statement.
At the absolute worst, the work of defining multiple, focused metapackages for the use cases of interest is the exact same amount of work as defining a "one size fits all" metapackage. But in reality, one size fits all is much harder because it's so nebulous. You say "a package that all active font maintainers in Debian can say they do at least "suggest"" as the end result; I think it's evident that that goal can't actually be delivered. And that's because both "universal agreement" and "covering all use cases" are un-hittable targets.
Individual metapackages _can_ be done. I'm saying start there, build a meta-meta-package out of the individual building blocks once you have them, and you might get closer to your goal.And, as an added bonus, even you personally don't care at all about the individual blocks (as it sounds thus far), they will serve many users' needs.
Nate
--