On Mon, Nov 18, 2002 at 11:40:20AM +0100, Sven Luther wrote: DB>> Sure, but it is not mandated in any kind of policy that they DB>> should be. And it can't be mandated, or such policy would soon DB>> become a mess with all the per-case clauses. I think "anything DB>> that gets executed goes to lib" makes things more clear cut. SL> Well, but the policy could let the choice to the package SL> maintainer, which should know if his package is arch indep or not.
Yes, this is also an option. SL>>> Finally, i don't know, but are they all that much people really SL>>> using /usr/share shared between different arches ? DB>> At least that is usage that FHS refers to, and the only practical DB>> effect of such division aside from abstract consistency. SL> Yes, but do you (or someone) know of someone really doing this ? No. SL> But again, is it really worth the trouble to set this up, only for SL> some hypothetical multi-arch installations needing this ? I agree on this one too. That is why I think that abstract consistency is more important goal for FHS than minor practical implications, and in this case the arch-dep/indep criteria serves more to compicate things than to make it more clear and consistent. -- Dmitry Borodaenko
