This one time, at band camp, Thomas Bushnell BSG said: > The point is that I want to massage some parts of the configuration > and not others. I want the others to continue to get updated by the > normal package installation process. > > If I use the one-big-file method, I can't really do this. I would > modify parts of the file, but then I can either install the new file > or not when the package updates it. It's a PITA, to merge every time > a small change is made in some other part of such a large config file. I wanted to mention that I completely agree with this sentiment. We run a couple of mail clusters, and we manage many single mail servers. Thanks to the split file approach, we can ship a couple of custom config files that allows us to tweak everything we need with a few macros and ifdef's. We still get the nice package management and improvement of routers and transports and all the rest of it. My only point in the previous mail is that there really is no difference in what is happening at run time - the only thing the end user has to know is that they when they pick one setting, they make edits here, and if they choose the other, they make edits there. -- ----------------------------------------------------------------- | ,''`. Stephen Gran | | : :' : sgran@debian.org | | `. `' Debian user, admin, and developer | | `- http://www.debian.org | -----------------------------------------------------------------
Attachment:
pgpKf_5o31IxP.pgp
Description: PGP signature