[Rpm-maint] --resign fails?
mh at novell.com
Fri Dec 15 16:44:18 UTC 2006
On Fri, 15 Dec 2006, 17:14:29 +0100, seth vidal wrote:
> On Fri, 2006-12-15 at 11:09 -0500, Bill Nottingham wrote:
> > seth vidal (skvidal at linux.duke.edu) said:
> > > That's fine - but then do we need to have a compiled tool to do that.
> > >
> > > my stated and brazen agenda is to put more and more things into higher
> > > level languages (python, ruby, perl, java, etc) above the rpm-lib so we
> > > can gain contributors other than just C programmers. Myself included.
> > Actually, the more I'm thinking - I think having the 'raw' from in the
> > executable helps; it's the lowest barrier to use in a recovery environment,
> > etc. For something like verification, you'd like to have a happy GUI thing
> > that shows you what's specifically changed rather than just a cryptic
> > set of flags, so *that's* the tool you build on top of the lib.
> by that argument do we have to have everything available in a compiled
> format for 'recovery' purposes?
> or can we count on python, for example, on fedora systems to be present?
I'd rather keep the 'recovery' functionality in the rpm executable
itself; python itself, and its runtime especially are known to
potentially create some portability issues which you don't want to deal
with in trouble situations... We should really fix the functionality in
rpm to deal with this (and similar stuff).
More information about the Rpm-maint