On Mon, Jan 28, 2013 at 07:39:25PM +0000, Dato Simó wrote: > On Mon, Jan 28, 2013 at 11:11 +0100, Iustin Pop <ius...@google.com> wrote: > > > > > + let lsock = optLuxi opts > > > > > + > > > > > + when (isNothing lsock) $ > > > > > + exitErr "Luxi socket (-L) required to execute jobs." > > > > > I haven't checked later patches to see how it looks, but if this is > > > > required, should be make a variant of oLuxiSocket that is enabled by > > > > default? Or will there be cases where you can run this tool without > > > > luxi? > > > > Ah. Yes, my initial plan was to give support for running without Luxi. > > > But (see my comments in the cover message) that resulted in somewhat > > > hairy code, so I decided against it. I'll send a 4.5/11 patch on Monday > > > to add such variant of oLuxiSocket if that's ok with you. > > > Yep, please do so. > > Actually, how would this work? I'm just not sure if you're suggesting: > > (a) a variant of the oLuxiSocket that somehow marks itself as required, > i.e. the user _must_ pass it > > or > > (b) a change in the default options in the harep personality, so that > Luxi is enabled by default. > > In the case of (a), I think I'm just missing how to mark an option as > required. In the case of (b) it seems that a non-trivial amount of work > would be needed so that the personalities can export an alternative set > of defaults to Main.hs. > > I'm okay with (a) if that can be easily done, but actually what may be > more sensible would be > > (c) just use oLuxiSocket as-is (to allow for alternate Luxi socket > locations), but have the code in Harep.hs _default_ itself to using > Path.defaultLuxiSocket. > > In (c), `harep` without any options works, but -L can be pass to > specify an alternate location of the socket. > > Thoughts?
I was thinking of a new option that is similar to oLuxiSocket, just defaulting to the defaultLuxiSocket, i.e. (b). But I'm not sure why this would be more work, instead of oLuxiSocket it would simply be oLuxiSocketAlways or such. However, (c) looks appealing, so let's go with that. thanks, iustin -- You received this message because you are subscribed to the Google Groups "ganeti-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to ganeti-devel+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
