Re: Keymap problems in D-I (was: Re: Bugs in the latest Debian Sid installer)
On Sunday 23 August 2009, Christian Perrier wrote:
> Quoting Frans Pop (email@example.com):
> > > I guess this is because the "apt-install console-setup" is in
> > > kbd-chooser's post-base-installer, at a moment where APT sources
> > > are not available (or configured).
> > No, it was simply because the package was not on the CD at all! And
> > that would have been a *very* obvious thing to check, don't you
> > think?
> Maybe, yes. If it was so obvious, why didn't you do it, then?
Because I was not handling the BR. You were. My point was that you stopped
half way in your analysis and tracing: you discovered that the bug was
specific to the netinst, but failed to make an effort to find out _why_.
> I understand *you* have a much better knowledge of D-I than most other
> people working on it right now. I understand (or rather I hear) you
> laughing silently in your corner when one of us is fighting stupidly
> with a bug the way I did with that one. I understand you think you
> would do a much better job than us right now and that, in some way, we
> suck at maintaining D-I.
No, I'm very much NOT laughing. I find it frustrating because of the fact
that console-setup has been in use for *months* and nobody who's been
pushing that change has ever found that issue or apparently even thought
whether the switch might also require change in debian-cd.
I'm crying because mostly because you IMO failed to make a decent effort
at tracing a relatively simple bug to the end.
[nonsense about communication skipped]
If you can't take a bit of mild criticism or ignore a mildly sarcastic
remark that's born out of frustration, then next time I'll definitely
just let you stumble around in the dark.
> > The rule is: if a package gets apt-installed, it *must* be included
> > on every bootable CD that includes the base system (i.e, all except
> > the businesscard).
> Point taken, though apt-install could very well install something *if*
> a mirror is available. But, yes, I understand this might hide problems
> like this one and we'd better avoid making this possible.
Yes it could, but *we don't want to rely on that*. We want to KNOW that an
image that contains the base system will always install without errors if
no external sources are used. That is how that image is defined!
And if you think about it for a few minutes it is not only logical, but
the *only* correct way to do it.