[bug#50358] [PATCH core-updates-frozen 0/8] Shortened Rust bootstrap & other fixes.

[Maxim Cournoyer <maxim.cournoyer@gmail.com>]

Hello,

Ludovic Courtès <ludo@gnu.org> writes:

> Hi,
>
> Maxim Cournoyer <maxim.cournoyer@gmail.com> skribis:
>
>> I feel like search-path-specification should be augmented to express
>> patterns that would be useful with search paths such as the too wide
>> XDG_DATA_DIRS: enable only when a child directory/file is present, for
>> example.
>>
>> So far we've been adding ad-hoc fixes in build systems (such as for the
>> qt-build-system, via (guix build qt-utils)); it seems it'd be cleaner to
>> add this capability at the search path level.
>>
>> What do you think?
>
> I’m of the kind who’d rather have reality match the model.  ;-)

Haha!  Tough!

> I think it would be nice to have a FONTCONFIG_FONT_PATH variable that
> would only look at share/fonts.

That'd be nice, but probably harder to convince them than for
XDG_DATA_DIRS, as we'd probably be the only user.

> What you describe would have uses beyond fontconfig, which is nice.  The
> downside is that it would make search path semantics and their
> implementation more complex; we’d have to see what the impact is.

Yes, more complexity, but localized (and can be tested), rather than
diffused (and untested) as of now.

> Thinking about it, I wonder if having (file-pattern "^fonts$") would
> work here.  But then, what if a profile contains several packages with
> an XDG_DATA_DIRS search path but different ‘file-pattern’ values?…

Perhaps this could work too, *if* the behavior of various same-named
environment variable specifications is to be combined into a logical OR
manner (I haven't checked what the current behavior is, but I seem to
recall it not working from past experiments).

Thanks,

Maxim