[pacman-dev] [arch-dev-public] Fwd: [arch] circular dependence in core

Xavier shiningxc at gmail.com
Thu Nov 1 13:38:21 EDT 2007


On Thu, Nov 01, 2007 at 06:17:31PM +0200, Roman Kyrylych wrote:
> 2007/11/1, Roman Kyrylych <roman.kyrylych at gmail.com>:
> > 2007/11/1, Aaron Griffin <aaronmgriffin at gmail.com>:
> > > Forwarding to the dev list in case people don't read the arch list.
> > >
> > > ---------- Forwarded message ----------
> > > From: Mister Dobalina <reebydobalina at yahoo.ca>
> > > Date: Nov 1, 2007 10:27 AM
> > > Subject: [arch] circular dependence in core
> > > To: General Discusson about Arch Linux <arch at archlinux.org>
> > >
> > >
> > > coreutils 6.9-3 -> depends on bash
> > >
> > > bash 3.2.025-1 -> depends on readline>=5.2
> > >
> > > readline 5.2-4 -> depends on ncurses
> > >
> > > ncurses 5.6-4 -> depends on coreutils -> oops!
> > >
> >
> > This was casused by this commit:
> > http://cvs.archlinux.org/cgi-bin/viewcvs.cgi/base/ncurses/PKGBUILD.diff?r1=1.24&r2=1.25&cvsroot=Core&only_with_tag=CURRENT
> > I think we should revert it (because dependencies of other packages
> > are fine and make more sense).
> >
> 
> And it was ncurses-5.6-4 package release that broke it.
> http://cvs.archlinux.org/cgi-bin/viewcvs.cgi/base/ncurses/PKGBUILD.diff?r1=text&tr1=1.24&r2=text&tr2=1.26&diff_format=h
> And we all missed this because pacman didn't complain when upgrading.
>

Well, no, this doesn't prevent pacman from installing the packages, only from
sorting them, which is usually not critical.
So pacman only prints a little warning in the --debug output when it detects
a loop while sorting the packages, but that's all.
Something like "dependency cycle detected"

I don't know if this message should always be printed, on standard pacman
output.
There could maybe be a pacman option or an external tool for detecting
dependencies cycles.




More information about the pacman-dev mailing list