This one time, at band camp, Robert Millan said: > On Sat, Jul 29, 2006 at 01:15:27PM +0100, Stephen Gran wrote: > > And then use constructs like > > > > if (which($postinst_hook)) { > > ... > > } > > > > I can't remember if you use taint mode and warnings and so forth, but if > > you do, substitute an array of known good paths for ENV{"PATH"}. > > I think it's better (asides from easier) to just not check that, and let > scripts fail if something is indeed broken in /etc/kernel-img.conf. > > Silently skipping broken commands in /etc/kernel-img.conf is not better than > failing and letting the user know about it, IMHO.
It currently silently skips them if they are not executable - I'm just providing a way to have the same behavior without full paths. Your patch changes the behavior of kernel-img handling in a backwards incompatible way. > > In any event, I am not sure this is RC, but I'll let manoj deal with > > that. > > It's release critical only on the basis that, without fixing this, a release > critical bug in another package (#361929) can't be properly addressed. RC bugginess is not necessarily transitive, at least IMHO. But as I said, this is for manoj to decide. -- ----------------------------------------------------------------- | ,''`. Stephen Gran | | : :' : [EMAIL PROTECTED] | | `. `' Debian user, admin, and developer | | `- http://www.debian.org | -----------------------------------------------------------------
signature.asc
Description: Digital signature