Jason Carley saieth:
> Sure there is room here for opinion. Mine however, as a new user
> of debian, is that dselect can be improved. It is not a slight on
> the fabric of debian merely an observation. Having just ben through
> the process of installation, I feel I can comment with some recency.
There is a lot of room for opinion. Many people feel there are serious
problems with dselects interface. People thought that when I was first
getting involved with Debian about 4 years ago. And most developers
accept that those problems and criticisms have merit.
This discussion unfortunately goes around in circles -- A new user,
such as yourself, starts using Debian, and raises valid complaints
about dselect's performance -- unaware that the issue is -very- old
(bug #1037, titled "dselect user interface (was Re: debian 0.95pl5
installation)" was filed 24 June 1995, and is still open), has been
discussed to death, and informing us of them isn't going to help. Old
timers reply by effectively saying "Deal with it", or even claiming
that it's good that it's so complicated. The new users (more than one
now) claim that this is the wrong attitude to have, and that the
problems shouldn't be ignored, etc. After a while, it boils down to
"Shut up and code".
The trouble is that no one has gone out and corrected those problems.
And very likely, for a variety of reasons, no one ever will. The new
users who have "shut up and coded" have looked at the code for dselect,
and have decided not to bother dealing with it. The old timers would
prefer to deal with something new (like apt or DPKGv2) rather than deal
with dselect. The code is some of the oldest for the project, and has
not had any major user interface changes in most of that time. It has
a reputation of being very fragile -- it will break if modified -- and
hard to work with. So most people realise that until something better
comes around, we are stuck with it.
"Shut up and code", followed by inactivity, isn't the only result of
the never-ending "dselect interface" debates. Some people have chosen
instead to work on dselect replacements -- replacements that don't
suffer from the maintainability problems that dselect has. The main
interface being worked on right now is apt. Currently, gnome-apt is a
decent, working packaging interface. In my opinion, it does not quite
replace dselect (mainly because I can't easily tell which packages are
new), but when it is done, it'll be a very good system. Apt is also
used as a back-end interface for dselect -- and here, it is superb.
the "apt" dselect method is a vast improvement in speed, flexibility,
and reliability over the older dselect methods.
> I am sure you are very experienced with debian and can thus use
> almost all of its tools better than I can. But there are areas
> where dselect can be difficult to follow and somewhat dangerous
> to use if you are not totally familiar with it. I guess it is an
> appropriate proving ground, a test if you will. But is that
> really the point of a package management application?
No, it isn't appropriate for a package management system to be a
"test", to make sure that only skilled or dedicated people install and
use the system. That isn't its intent, however good it may be at it.
And yes, there are dangerous, poorly documented areas of dselect. I
still remember the first time I installed Debian -- there were several
packages I wanted to get rid of after the default install, so I went
into the "select" option, cleared all my selections, chose the few
packages I wanted to remove, fought my way past the dependency screen,
and proceeded to [R]emove almost my entire system. Shortly thereafter,
I installed Debian for the second time ;-).
But until someone writes something better, or fixes dselect, it's what
we have.
--
Buddha Buck [EMAIL PROTECTED]
"Just as the strength of the Internet is chaos, so the strength of our
liberty depends upon the chaos and cacaphony of the unfettered speech
the First Amendment protects." -- A.L.A. v. U.S. Dept. of Justice
