[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#31449: Debian Bug report logs - #31449

Martin Schulze <joey@finlandia.Infodrom.North.DE> writes:

[Adam wrote]
> > I already had my system installed on the disk (which was mounted as
> > /target), and it was running 2.0.35 rather than 2.0.36, thus
> > the messages.  Of course another minor problem was that the error
> > messages weren't redirected to ttyp3.

> > But the major issue is that dbootstrap didn't detect that I was
> > trying to install on a system which wasn't in an appropriate
> > state.  I know that one is not supposed to do that; still, it
> > should catch that condition.

> What do you expect?

I think, yes, some sanity checks in dbootstrap is going to make life
easier for users and fewer questions and bugs for us.

> Define the sanity checks you'll want to see.

Yes, that's the rub.  Sanity checks generally break more often than
the stuff they are "protecting".

> The only thing that I could think of in this case would be dbootstrap
> checking if /target/lib/modules/`uname -r` exists and refuse to let
> you start modconf if it doesn't.

That's a good one.

> I could also think of some sanity checks that will prevent dbootstrap
> from extracting the base system if some random files already exist.

Yes.  I think perhaps the moment this should be done is when existing
filesystems are mounted.

Some checks would be, for instance, /etc/debian_version, uh, likewise
a file for RedHat ?  Anyhow, such checks should result in a warning,
allowing the user to go on if they want to.

.....Adam Di Carlo....adam@onShore.com.....<URL:http://www.onShore.com/>

Reply to: