Encouraging the use of multiple packaging systems on one systems, and the resulting problems (was: re: /usr/local)

[Konstantin Ryabitsev]

On 21/10/05, Mike Hearn <mike at plan99.net> wrote:
> > People using the autopackage - they must find out by some other means that
> > an update is available to fix a problem they may not know exists.
>
> Don't panic, it's just a bug. We have quite detailed blueprints for
> an automatic online update service, it just hasn't been implemented yet.

Autopackage has been "solving the world hunger" for the past few
years. It has failed so far (I've seen a bonafide .package file
offered to me ONCE in my entire life), and there are no indications
that the "tides" will turn. Nobody is interested in a Packaging and
Updating System to Obsolete All Packaging and Updating Systems, since
every major distribution has already moved on past this problem:

* With Debian/Ubuntu you can already install almost any bit of
software ever written
* Fedora Extras is getting us there
* Gentoo et al are also not interested
* Linspire has their own system
* Proprietary vendors just release statically compiled tarballs, or
generally lack clue (usually both)

Why are you so set on making it more difficult to support the systems?
In our previous conversation on the Ubuntu list, I've already shown
you how Autopackage will lead to problems, since its dependency
tracking only checks for things that are required by the package
currently installed:

* foo requires libbaz-1.0, which autopackage fetches and installs
* bar requires libbaz-2.0, which autopackage fetches and installs
* foo stops working, since its dependency is no longer there

If your answer starts with "b" and ends with "undle", then how is this
different from LSB RPMs and the aforementioned statically-happy
vendors?

Can you state clearly the problem that autopackage is trying to solve?
As a bonus, can you state clearly how autopackage won't ultimately do
more harm by potentially breaking systems or leaving them vulnerable,
since it won't do updates? And when it does become smart enough to do
updates, can you tell us how your claim that "repositories are not
sustainable" will still hold true?

It seems to me that you're repeatedly trying to enter a door that's no
longer there.

--
Konstantin Ryabitsev
http://www.mricon.com/