On Tue, Apr 29, 2014 at 09:22:26PM -0400, Sam Hartman wrote: > I'll withdraw my objection to krb5-config moving to krb5-multidev. > Shouldh we use alternatives for managing krb5-config from Heimdal vs > MIT? Yeah, that would be useful. MIT should probably have a slightly higher priority than Heimdal so that it gets used when both are installed.
> For what it's worth, I don't support asking package maintainers to > update build-depends. The reason I thought introducing krb5-multidev > made sense was for the few packages that need both installed at the same > time. The reason I'm interested in having dependencies on the -multidev packages (besides building packages that depend on both heimdal and mit) is because it means not having to regularly install and uninstall a bunch of packages to work on different software. One of the codebases I work on depends on libsvn-dev. libsvn-dev depends on libaprutil1-dev, which depends on libpq-dev, which depends on libkrb5-dev (now krb5-multidev). Every time heimdal-dev is installed, it ends up kicking out libkrb5-dev and all the stuff that depends on it. I mostly care about having the -dev package depend on krb5-multidev, rather than the build-depends. The latter is only somewhat useful because it means being able to build the source package while having heimdal-dev installed. > I think people who want to do so can file wishlist bugs saying that > using -multidev packages and krb5-config.mit|heimdal has certain > advantages. > However it has complexity disadvantages (admittedly significantly fewer > once we move krb5-config) and I think it should be viewed as maintainers > MAY use krb5-multidev|heimdal-multidev rather than maintainers SHOULD. You're right, I should have filed those as wishlist bugs. Avoiding conflicts in -dev packages is probably not big enough of an issue to warrant a "normal" bug. Cheers, Jelmer -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
