On Thu, 10 Jan 2002, Dimitri Maziuk wrote: > * 'cduck' Chris Grierson ([EMAIL PROTECTED]) spake thusly: > ... > > > > i think you misunderstood my intention. i wasn't trying to veil a > > complaint, > > rather ascertain the proper way to mass-install a large list of packages w/o > > encountering the problem of not packages install various files/links/dirs > > because a package has not yet been installed (even if they are all installed > > via the *same*, *single* call to 'dpkg -i *.deb'). > > "apt-get upgrade" (with "dpkg --[get|set]-selections") is probably safer than > "dpkg -i *.deb". One of the things apt was supposed to do is install packages > in proper order. It's been a while since I looked closely at Debian packaging > system, perhaps dpkg does that too nowadays... (and BTW aren't you supposed to > use "-iGREB" instead of just "-i" when you use dpkg that way?)
ok, this is what i was looking for--you are supposed to use dpkg with --get-selections on one box, input that file to 'dpkg --set-selections' on another, and then use 'apt-get dselect-upgrade' (apt-get upgrade will not honor the work of 'dpkg --set-selections')? if that is the proper way to do things, that solves one problem, but i believe that part is only aesthetic; 'dpkg *.deb' will install packages in the correct order *so as to satisfy dependencies*, but that isn't the issue. the issue is whether the 'preinst' scripts should be checking for installation status of a package by means of either 'installed' or 'unpacked', meaning that it is about to be installed, just like the one running the script is. in the case of doing something like 'dpkg -i j2sdk1.3*.deb mozilla*.deb', and j2sdk1.3 gets installed first (since neither depend on each other, and j2sdk1.3 is specified first *and* for good measure, it's lexically first), it won't install a crucial link to get mozilla to use the java plugin. is this an inherent issue with the package scripts, or with batch-/ mass-installing packages or ??? > > Dima > -- > "Mirrors and copulation are abominable because they increase the number of > entities." -- corollary to Occam's > Razor thanks, -c [ Structural Informatics Group ] [ Dept. of Biological Structure ] [ University of Washington ] [ 206.616.7356:office ] [ 206.795.4998:cell ] ------------------------------------------------------------------------
