On Thu, Jan 12, 2006 at 05:17:57AM +0800, Syan Tan wrote: > in bootstrap-local-first.conf I have to put > the path to plpgsql.so in the language library dirs$ > to make it work, Why ? I need hard facts to believe that. Why would it not work to put it into *monolithic* ?
> but apparently this is supposed to be empty, > and admins have to change it first , if they know > the bootstrapper isn't going to find plpgsql.so ; local_first is intended to give local admins a chance to have arbitrary things being done to the newly created database before the standard GNUmed bootstrapper starts doing it's stuff. This is standard practice in config files, for example. Now, if a local admin has to put things into local_first/last, fine, that's what they are for. However, those changes are NOT to be channeled back into CVS exactly BECAUSE of the local nature of the changes. If the changes would be put into CVS they would raise conflicts with local changes at OTHER sites. That'd be missing the point completely. > and they may only know this by trying to boostrap , > having the boostrap fail The bootstrapping shouldn't fail. If it does it's our fault and there's a bug (which may be hard to tackle, though). The local_first/last files are not supposed to make any difference to the bootstrapping proper. Karsten -- GPG key ID E4071346 @ wwwkeys.pgp.net E167 67FD A291 2BEA 73BD 4537 78B9 A9F9 E407 1346 _______________________________________________ Gnumed-devel mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnumed-devel
