Toralf Lund <toralf@xxxxxxxxxxxxxx> on 11/02/2005 09:04 AM added:
1. 'rpm -U' is equivalent to 'rpm -i' on the package files specified,
followed by 'rpm -e' of any older versions of those packages that
are already installed.
Not wanting to argue about it, but shouldn't that actually instead be
the other way around?:
rpm -e oldversion followed by
rpm -i newversion
because... if both versions contained file 'A',
wouldn't that mean that 'A' now contains the new code,
and then because the erase happened second,
that 'A' would then be erased/
It is quite definitely "install first, then erase" for -U. Both
sequences have their pros and cons, but I believe the actual choice is
based on the argument that if you were to remove packages first, then
files being replaced would be missing completely for a very short while,
and might be lost from the system if the operation failed halfway
through - and there may be files that must *always* exist in the system.
Also, I believe the erase operation is smarter than what you suggest
above; it won't ever remove a file that's still referenced by a package.
In other words, if you install the same file twice via rpm -i (with
--replacefiles) of two packages that contain (different versions of) the
file, then do rpm -e on one of the packages, the file will not be
deleted. Which version of the file you are left with depends on the
install sequence, I think.
- T
_______________________________________________
Rpm-list mailing list
Rpm-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/rpm-list
