Hi Nathan,
Am Dienstag, den 05.02.2019, 13:45 +0000 schrieb Nathan Willis:
> (a) only makes sense within a limited context -- namely, the type of
I think Adam has made quite clear in his initial announcement what the
objectives of this meta-package will be. It is about (a) providing a
reasonable set of fonts to start with (think of Serif, Sans and Mono in
all of Regular, Bold, Italic and Bold-Italic), so you have enough to
begin a heading/body type of text, (b) to be compatible with other
operating systems (especially Windows standard fonts) and (c) to have a
reasonably wide Unicode glyph coverage.
If that is the case, then I am contending that there _is_ a context here -- but that it has to be identified and made explicit in order for the need to be met successfully.
Consider you are installing e.g. Windows 10. It will come with a lot of
fonts that serve just that purpose.
Ah, but it's purpose_s_, plural.
Some standard fonts for word
processing, some legacy fonts for compatibility with older OS versions,
some purel decorative fonts and some more fonts to cover even the most
exotic glyphs. Someone at Microsoft has decided that this is a good
choice of fonts to have on an average system and we want to provide the
same service with the fonts-recommended meta-package.
Here, for example, you've listed four different contexts. And I think (c) from the preceding paragraph above that is a fifth. [In (a) it seems like you're describing the same thing as "word processing" below, and (b) is definitely the same as the OS-compatibility one in the second paragraph.]
I'm saying: that's five metapackages. Not one that does-everything-possible. Having 'fonts-wordprocessing' and 'fonts-os-compatibility' both available (etc.) is a better experience than trying to roll them all into one.
Easier for the user to decide that it is desirable, easier for the packagers to scrutinize & choose, and easier to maintain over the long run.
dependencies, it converges towards a "universally accepted
recommendation" if something like this exists.
Indeed, "universally accepted" for all use-cases is what I am saying isn't doable. Being more focused, however, with multiple metapackages, untangles that knot.
I know that someone asking me "what fonts should I install for full Unicode coverage" is drastically easier to answer defensibly and with confidence than "what fonts should I use" in the abstract.
Thanks,