Dne 21.12.2016 v 18:08 Matthew Miller napsal(a):
On Wed, Dec 21, 2016 at 11:18:54AM -0500, Randy Barlow wrote:
>>> to move it to /usr/lib/fontconfig/cache.
>> This proposal likely is incompatible to the FHS.
>>
>> Files in directories below /usr/lib are supposed to static and not
>> to be dynamically created.
> I agree - let's not violate FHS.
Keeping to standards where they make sense is valuable. In this case, I
want it to be clear that this isn't just a "doesn't meet red tape!
denied!" stamp — the desire is to avoid runtime-generated files under
/usr, right?
While you are afraid, I see in the bugzilla a lot of "it was done
previously, we can do it again". And next time this case will be used as
a justification for moving more and more stuff into /var.
When I look at the details, this seems like an edge case. See the
linked bug (
https://bugzilla.redhat.com/show_bug.cgi?id=1377367). This
isn't really _runtime_ variable content. It's generated at
package install time. The FHS says:
/var is specified here in order to make it possible to mount /usr
read-only. Everything that once went into /usr that is written to
during system operation (as opposed to installation and software
maintenance) must be in /var.
... and the thing is, this particular cache is generated during
"installation and software maintenance".
I see that the fc-cache data is arch-specific, but is it really
host-specific? What if we just pre-generated it at *build* time and put
it into /usr/share or /usr/lib? I guess that'd end up making font files
archful, but that's not the worst thing.
I live under impression (may be wrong) that it should be safe to delete
content of /var/cache on disk restricted devices ...
Vít
Hmmm, I see that the arch-specific part is encoded in the filename
(like d3379abda271c4acd2ad0c01f565d0b0-le64.cache-7). None of this
stuff is very big; only a few k per font. What if we just included all
of the arch-specific files in every font package?
Bonus: less stuff run in %post scripts!