|
From: <las...@Su...> - 2001-07-09 12:09:00
|
Dan Mueth wrote: > = > On Thu, 28 Jun 2001, L=E1szl=F3 Kov=E1cs wrote: > = > > One more thing, with this we would have three ways of specifying OMF > > file locations. > > > > We have: > > > > - and environment variable OMF_DIR > > - the scrollkeeper.conf variable OMF_DIR > > - the -o flag of scrollkeeper-update > > > > This is probably too much. I think the environment variable should go= > > away and maybe the -o flag also. If -o stays then the final set of > > directories should be the unification of the -o list and the OMF_DIR > > list from scrollkeeper.conf. > > > > Thoughts? > = > We certainly need to keep scrollkeeper.conf, so distributions can set u= p a > reasonable system. The OMF_DIR variable is very nice for people > developers, who set up different paths with development or stable > versions. It is also the only nice way for users to set up their own > databases in their home directory. I suppose you could have a > .scrollkeeper file in a user's home directory instead of having OMF_DIR= , > but that doesn't really work for systems where you want to switch back = and > forth between using two different databases (as done by developers with= > stable and unstable prefixes). So, I think we really should keep the > environment variable too. > = > I believe we originally introduced the -o flag so that when a person is= > building an RPM they create a database in a path under the BUILDROOT. = In > this case, using scrollkeeper.conf would do the wrong thing. I'm not > sure if we can simply set OMF_DIR before running scrollkeeper-update. = It > seems like keeping the -o flag is a little more elegant. > = > So, I think we should keep all three of these. > = > Dan Ok, sounds good. Laszlo |