On Oct 29, 2010, at 10:53 AM, devzero2000 wrote:
> Sure. It was only an little observation. Sorry if not strictly related.
Getting pseudo wired into rpmio, and/or using mount -bind, or
just handling the +/- chroot prefix directly in RPM (all the wrapped
syscall intercepts exist for "RPM ACID" transactional logging) and
otherwise permitting RPM to be fully featured for non-root oddly
skew RPM development.
E.g. last night the bug in rpm-18.104.22.168 preserving CWD across
internal lua scripts (in the RHEL6 setup package) when chroot(2)
is performed bit me Yet Again.
I really wish these bugs could just be "fixed", not "faked". There are
design flaws in using chroot(2) (which requires root to use) with
tools to be used by non-root. Sure there's endless hack-o-rounds
to solve the issues.
73 de Jeff
Received on Fri Oct 29 17:18:38 2010