On Mon, Apr 18, 2016 at 01:14:43PM -0600, Bob Proulx wrote: > Florian Lohoff wrote: > > After diving into the problem the cause was that actually not the > > shipped sa-update was called but a locally (and broken) version > > in /usr/local/sbin/ > > > > The postinst should take care of not calling binarys with relative > > path or relying on the PATH variable: > > No. The entire purpose of the /usr/local hierarchy is to provide a > way for the local admin to override and install their own software. > With great power comes great responsibility. If something the local > admin does there causes breakage then it is their responsibility to > fix it there too.
Okay - then close this bug ... The problem i had that package installation failed for completely unrelated problems in files somewhere else on the system and this was a difficult thing to actually find and correct. It brought the system to a point where the admin had no clue on how to proceed with system upgrade and was in the decision to reinstall the whole system because he failed to find the exact problem. I would have expected package installation to be VERY self contained and actually this was the very first time in 20+ Years that something like this happened to me. I have never heard of overriding binary in postinst which stem from THAT very package was actually a good idea. Flo -- Florian Lohoff [email protected] We need to self-defend - GnuPG/PGP enable your email today!
signature.asc
Description: Digital signature
