Re: [custom] Custom Debian Distros need the help from debian developers
On 10-Mar-04, 07:49 (CST), cobaco <email@example.com> wrote:
> The argument against using debconf this way is usually that this would
> present the user with to many questions on installation of the package,
> though as Petter explains in  you could use hidden debconf questions
> which would alleviate the problem.
No, the argument against using debconf that way is that for many
packages, there's no need for debconf *at* *all*: they can ship a
working conffile, which is handled automatically by dpkg w.r.t. upgrades
and preserving local changes. As soon as you use debconf (even if
the questions aren't displayed), you can no longer use the conffile
mechanism, which is a giant step backwards.
I can't begin to count the number of times debconf using packages have
overwritten local changes. Redhat at leasts saves a backup!
Now, I'll be the first to say that this isn't debconf's fault. It's
entirely due to incorrect configuration file handling with the offending
package. But the wide-spread occurence of this offense suggests that
most maintainers *don't* understand how to use debconf (despite correct
examples), and don't really understand the "debconf is not a registry"
And no, questions of the form "Allow debconf to manage this
configuration file?" are no the correct solution, and IMO, violate
Debian Policy. But I know others disagree.
I suspect that proper use of ucf solves the problem, and gives
conffile-like handling to non-conffiles. But until use of ucf for
"debconf-ized" packages is mandated, it won't be sufficient.
The irony is that Bill Gates claims to be making a stable operating
system and Linus Torvalds claims to be trying to take over the
world. -- seen on the net