On 10/9/07, Casper.Dik at sun.com <Casper.Dik at sun.com> wrote:
>
> >Absolutely. The package tools could do with a bit of loving
> >care and attention, but aren't fundamentally broken and
> >there's no reason why they need to be replaced by any
> >other solution.
>
> I'd love to do a small side project which make the package tools
> much faster by optimizing the way /var/sadm/install/contents
> is written (not hard to write the mechanism but a bit daunting to
> put it inside the package tools)

It would be interesting to do that. And interesting to see how much
difference it makes. It ought to make quite a bit of difference, but
actual measurement indicates that there are other issues that are
even worse.

There are a couple of things that I would like to get done for SVR4
packaging. One is to eliminate SVR3 support - it seems unlikely that
the odd fragments of code present can possibly do anything useful.
The other is to refactor the code which has grown rather untidy.

In terms of actual functionality, the two items that sit at the top of my
list are (i) correctly re-ordering packages supplied to pkgadd or
pkgrm so that they get added or removed in dependency order (dealing
with dependencies outside the argument list is out of scope, but if you
tell it to add a couple of packages then it should be capable of working out
the correct order to do so), and (ii) creating something that allows you to
replace a package by another version of the same package in a single
operation.

-- 
-Peter Tribble
http://www.petertribble.co.uk/ - http://ptribble.blogspot.com/

Reply via email to