On 20 July 2011 02:21, Ionut Biru <ibiru@xxxxxxxxxxxxx> wrote: > I'm concern about this replace and I don't know if pacman can handle this > well enough. Does pacman accept a versioned replace to not > conflict with the newly virtual-guest-additions? > > replaces=(virtualbox-guest-additions<4.1) ? > > I don't want to hit this case: > > virtualbox-additions replaced by virtual-guest-addtions 4.1.0 > virtualbox-guest-addtions 4.0.12 replaced by virtualbox-archlinux-addtions > 4.1.0 > virtualbox-guest-additions 4.1.0 replaced by virtualbox-archlinux-addtions > 4.1.0 I remember now. You expressed this once in IRC. You will have to work around this or post an announcement to warn. I believe this is the same situation as with Python 2.x and 3.x modules. There is no versioned replaces in pacman yet [1]. If it all seems a bit too much work (it does actually after I look at it again), then I suppose simply getting rid of the 'guest' would do. In that case, it warrants a different scheme: virtualbox-additions virtualbox-(arch)linux-additions || virtualbox-additions-(arch)linux virtualbox-(arch)linux-modules || virtualbox-additions-(arch)linux So, you don't rename the main guest additions package. With the appropriate/relevant descriptions, IMO, this is the next best KISS. Also, I don't think it's the main package that's the source of the confusion/criticism, but the linux-specific ones. In this case we would be renaming only the linux-specific packages, so the ultimate purpose would be served. [1] https://bugs.archlinux.org/task/23410 -- GPG/PGP ID: 8AADBB10
