Bug#840146: pbuilder: remove_packages checks for existence in host, not in chroot

[Thorsten Glaser]

James Clarke dixit:

>I agree, this is bad. However, if this is only affecting sarge chroots,

I never said that. That’s just where I noticed it.

Every time anything listed in REMOVEPACKAGES has a different state
between the host system and the chroot, and when the chroot’s distro
doesn’t actually have that package at all, this makes --update fail.

>If it's problematic for wheezy+ chroots, please feel free to change
>the severity back to serious.

It’s about 50% likely to do that.

I can’t construct a realistic case right now, but I think if you
add libeatmydata1 to REMOVEPACKAGES, an --update of a wheezy sans
backports chroot will fail.

Honestly, just quickly upload a fix and let’s not hang ourselves
up on the severity.

Thanks,
//mirabilos
-- 
<igli> exceptions: a truly awful implementation of quite a nice idea.
<igli> just about the worst way you could do something like that, afaic.
<igli> it's like anti-design.  <mirabilos> that too… may I quote you on that?
<igli> sure, tho i doubt anyone will listen ;)