Re: Switching the default startup method
* Raphael Geissert (geissert@debian.org) [090824 00:34]:
> Andreas Barth wrote:
>
> > The local admin doesn't have any
> > choice. I need to admit that I disagree with this change at this
> > time.
> > #475478 insserv: uninstallation fails horribly if an init script has
> > been removed.
> [...]
> >
> > #538959 needs actually to be worked on. The current state is not how
> > it should be.
> (which you later said it should be #511753)
>
> These two only seem to occur when insserv is removed, and since it is now
> pseudo-essential that should never happen, IOW: IMO they are not grave (I
> do agree that they should be addressed, though).
Eh. This translates to: "it is ok that the admin cannot switch back
from insserv to oldstyle booting".
And that is a statement that I heavily disagree with. I think neither
our users nor our developers at large considers that a feature, but
rather a very grave bug.
We should definitly continue to support oldstyle booting, at least for
the time being.
> > #538959 is really quite serious. We have cluebatted maintainers for
> > quite some time to use update-rc.d instead of other methods. Breaking
> > this by insserv doesn't get you support. So please fix this ASAP, but
> > without breaking the traditional methods even more.
> >
> > Also I do admit that I'm a great fan of enabling our users to make a
> > choice. For lots of server based systems having the oldstyle sysvinit
> > scripts works very well. It's easy to understand, it's obvious, it's
> > standard since quite many years. I don't see any reason to enforce
> > dependencies on these people, and as you can see in #538959 quite many
> > people don't want it.
>
> I think the missing point here is that insserv is just one of the ways to
> fix the problem of having to guess a correct start number, among many
> others; and any system that doesn't implement that is actually a
> regression. There are other tools similar to insserv that also do
> dependency-based booting (but AFAIK none of them are in Debian).
So you are telling us here that anyone who depends on the 20+ years
working method of ordering boot with decimal numbers is using a
regression? Sorry, but this is just plainly wrong. It is a regression
if you change stuff in a way that it breaks a working scenario for
many people.
Please do not break something that works. Please be conservative with
changes to essential stuff. Debian is not a sandbox nor your
playground, but used on many production systems. The trust that we do
mainly sane decisions is one of our most valuable possesions. And we
should make sure we keep that trust.
Cheers,
Andi
Reply to: