On Tue, Apr 8, 2008 at 10:46 AM, Bart Smaalders <[EMAIL PROTECTED]> wrote: > Danek Duvall wrote: > > On Tue, Apr 08, 2008 at 08:15:35AM -0500, Mike Gerdts wrote: > >> The same interface may be provided by different packages. For > >> example, postfix and sendmail both provide an executable called > >> /usr/lib/sendmail and speak SMTP on port 25. For those people that > >> aren't doing a lot of uucp mail relaying and like a human readable > >> (and writable) configuration, postfix is a very attractive software > >> package. > > > > It strikes me that depending on the entire OS just to pull in SMTP server > > capability is a bit broad of a net. Wouldn't this be an appropriate place > > to use your "feature" category? > > > > Both approaches are needed. Being able to evaluate your package's > dependencies by specifying an incorporation of all the system's packages > is clearly useful as a way of saying "evaluate this package against > build 79"; > at the same time, having the publishing software record those > dependencies as > dependencies against individual packages is very useful for allowing > minimization. > > It's not clear to me as to best handle the sendmail/postfix dependency. > Perhaps > the cleanest approach is the mark both sendmail and postfix packages as > providing > svc:/networking/smtp and requiring this in the dependent package. Such > choices would of course > prevent automatic installation of dependencies as an informed user > selection would be required.
I don't see why the system couldn't be distributed with a default set of choices with the caveat that the administrator must audit those to ensure that their desired automated configuration will be enforced. One of the things I hate the most about Debian is the questions it asks me everytime during large package upgrades or installations (since inevitably, each time I do so, it is a new system I'm doing it on). I still believe generic package dependencies like these are problematic since they depend on each packager providing a generic dependency that interprets the requirements of that dependency in the same way. I also believe that there needs to be a way for packages to explicitly state that you must have MySQL *or* postgre as opposed to just "database" (which is common among certain programs, libraries, etc.). Somehow the system then also needs to record the choice that was made so that when an attempt to remove MySQL is made (even if postgre is installed) a warning can be triggered because the package was originally installed when only MySQL was present. -- Shawn Walker "To err is human -- and to blame it on a computer is even more so." - Robert Orben _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
