On 6/19/06, Bo Ørsted Andresen <[EMAIL PROTECTED]> wrote:
On Monday 19 June 2006 10:58, Trenton Adams wrote:
> > > One other benefit of this mechanism, that I can think of, would be
> > > that I could easily roll back to my last sync date or release, if some
> > > of the updated ebuilds caused me problems.
> >
> > That really should not be necessary. What you should of course do is file
> > a bug so the problems can be fixed for everyone.
>
> Well, one problem I had was not actually *really* bug. It was a
> requirement that I did not fulfill, but was unable to figure it out
> instantly. So, rolling back would have been very useful at that time.
> It would just add another level of safety.
If an ebuild is removed from the tree while you still need it then chances are
that others need it too. Then it is indeed a bug. If you need an ebuild that
has been removed from the tree it is available from the cvs [1]. As mentioned
in my previous mail cvs is going to be replaced by a superior VCS hopefully
within this year.
Sorry, you misunderstood what I was saying because I wasn't clear
enough. I meant if I missed a required step, not missed a dependency.
For example, with openldap, you're supposed to slapcat before
upgrading, and slapadd after upgrading, or you could have database
problems. I did not know this. So when I went from 2.2-2.3, I had
problems. But, I didn't have time to get it working, as I needed it
up and running NOW. So, I reverted to the old package by masking the
new one, and then went to find out why it occurred after the fact.
This is one simple example of potential problems. But something on a
wider scale could occur.
[1] http://www.gentoo.org/cgi-bin/viewcvs.cgi
--
Bo Andresen
--
gentoo-user@gentoo.org mailing list
