2007/11/18, Xavier <[EMAIL PROTECTED]>: > On Sun, Nov 18, 2007 at 11:42:11AM +0200, Roman Kyrylych wrote: > > Huh? > > gcc-libs doesn't depend on gcc > > python doesn't depend on tk > > I guess makedepends were counted here, which is wrong. > > > > Ah indeed, the script doesn't make any distinction between makedepends and > depends. It probably should just ignore makedepends for the cycle check. > I made it ignore makedepends totally, just for checking which cycles are left > (but the script should keep checking the makedepends exist). > > That reduces the number from 26 to 13. And there are several duplicates there > also, so I regrouped them manually (that should be done in the script > though;)), > and we are left with 8 : > > coreutils>pam>db>coreutils > > shadow>pam>db>coreutils>shadow
> pam>db>coreutils>pam > db>coreutils>pam>db Why the hell empty functions with /bin/true are still in db.install? This commit removed some of them: http://cvs.archlinux.org/cgi-bin/viewcvs.cgi/base/db/db.install.diff?r1=1.1&r2=1.2&cvsroot=Core&only_with_tag=CURRENT Removing others will eliminate the silly dependency on coreutils. I guess there are much more .install scriptlets with just /bin/true in functions. I think we should modify install.proto to not endorse this bad behaviour. > > tzdata>glibc>tzdata > glibc>tzdata>glibc I think tzdata should not depend on glibc. -- Roman Kyrylych (Роман Кирилич) _______________________________________________ arch mailing list arch@archlinux.org http://archlinux.org/mailman/listinfo/arch
