On Fri, Mar 11, 2011 at 04:23:03PM +0000, Adeodato Simo wrote: > Hi, > > I don't have a real opinion on this topic, but I wanted to share my view > on a couple points: > > On Fri, Mar 11, 2011 at 15:07 +0100, Iustin Pop <ius...@google.com> wrote: > > - people will need the Haskell platform in order to build ganeti (could > > be worked around by making htools optional, but I dislike this option) > > It's pretty standard (and customary) in the OSS world to allow for > optional compilation of modules via ./configure. > > For this reason, I would really recommend allowing to skip compilation / > installation of htools via a ./configure switch, even if the default is > to perform such compilation. > > I think it can also be beneficial for the project: if a user is trying > Ganeti for the very first, not making the barrier for a "first set-up" > higher might result in one or two users not being discouraged away. ;-)
True. OTOH we do require a non-trivial number of Python modules, so it's not like we're a very "plug-and-play" setup. We'll have to come up with a reasonable way to disable features; note that our current Python code doesn't really have optional features that reduce the number of dependencies. > > > - ganeti needs actual compilation, so it's not Arch: all anymore (in > > > Debian speak) > > > The main package can still be Arch:all and you just build a separate > > package for htools. > > This is a standard thing to do too, as you probably know. To clarify (for the 2nd time): my problem is that ganeti (the source package) will _also_ have arch:any packages. Having just arch:all packages is nicer, because you don't have to deal with porter boxes. It's not a big deal, but it is something to keep in mind. thanks, iustin
