'Twas brillig, and Thierry Vignaud at 31/03/13 17:30 did gyre and gimble: > On 31 March 2013 18:22, Thierry Vignaud <[email protected]> wrote: >>> Preparing... ##################################### >>> 1/154: dash-static ##########################warning: >>> group daemon does not exist - using root >>> warning: group mail does not exist - using root >>> ########### >>> 2/154: glibc ##################################### >>> 3/154: lib64termcap2 ##################################### >>> 4/154: perl-base ##################################### >>> 5/154: update-alternatives ##################################### >>> 6/154: bash ##################################### >>> 7/154: lib64python2.7 ##################################### >>> -/154: ##################################### >>> >>> >>> It seems the last one in that list is filesystem, but it has neither a >>> number, nor is it's name printed. >>> >>> I suspect this might related to the fact that they key is no present >>> which I'm attributing to the fact this is a blank chroot rather than an >>> actual signing problem. >> >> It means we failed to get this package fullname >> http://svnweb.mageia.org/soft/rpm/urpmi/trunk/urpm/install.pm?revision=7595&view=markup#l121 >> >> After debuging, callback_open got called one too much (it got called >> twice for first package >> which is actually filesystem), thus we bump the index one too much. >> Sadly the doble call came from librpm's rpmtsRun() >> >> Interestingly, the open & close calls are just before and after rpm prints: >> "warning: filesystem-2.1.9-19.mga3.x86_64: Header V3 RSA/SHA1 >> Signature, key ID 80420f66: NOKEY > > It happens b/c filesystem as a %pretrans > runTransScripts() calls rpmteProcess() which says: > /* Dont bother opening for elements without pre/posttrans scripts */ > > But since we now have package with %pretrans due to /usr migration, > it goes further and calls rpmteOpen()->rpmteFDHeader() > which notifys us about a package opening in order to get its file descriptor. > > Since we now have %pretrans scriptlets, I'll adapt urpmi in order not to > bump index out of actual transaction.
Awesome :) Good debugging and interesting info above! Thanks as always Thierry :) Col -- Colin Guthrie colin(at)mageia.org http://colin.guthr.ie/ Day Job: Tribalogic Limited http://www.tribalogic.net/ Open Source: Mageia Contributor http://www.mageia.org/ PulseAudio Hacker http://www.pulseaudio.org/ Trac Hacker http://trac.edgewall.org/
