2014-03-31 15:29 GMT+02:00 "Jóhann B. Guðmundsson" <johannbg@xxxxxxxxx>:
> You do realize that the size of the base needs to be bound to the lowest
> common denominator between current and future multiple products to make it
> this kind of proposal work now and in the future right?
We are at liberty to force all possible products to carry some
overhead to simplify our work or user's expectations. For example,
even if no program in Fedora ever called glibc's hcreate() or
strfry(), we wouldn't consider dropping the implementation and
breaking the ABI; and the same thing can happen at the library level:
We are quite at liberty to promise that libvirt.so.0 will be available
on every Fedora system (even if any attempt to connect to a hypervisor
returned an error.)
Sure, such choices to simplify our environment at the cost of Base
size would disqualify us from the "smallest distribution for embedded
use", but we have never really been in that race anyway, and I'm not
even sure that size matters all that much for the embedded cases where
Fedora is an option (compared to ease of deployment, or availability
of "common" APIs, for example).
> since by doing
> so they invalidate their own logical approach of building and limit baseWG
> to self hosting components.
Only creating Base from self-hosting components does not actually make
sense to me: fedup is not necessary to compile any package, and yet it
clearly belongs in base; various specialized build tools like
documentation extractors and build test frameworks, or even make(1),
are necessary to build packages, but we may not want to promise their
existence to Base users.
Mirek
--
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxxx
https://admin.fedoraproject.org/mailman/listinfo/devel
Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
