Hi people!

First of all - I really do read all these posts and try to think hard
about them. But right now it's weekend and I don't have time to answer
until tomorrow.

Here is a quick one though.

Julian Fitzell <julian at beta4.com> wrote:
> lex at cc.gatech.edu wrote:
> > Sure, keeping the information is fine, and there are probably exciting
> > things you can do with it.  Just don't use it as dependencies.
> 
> Goran's plan has always been to use the information (combined with an 
> indication on a release of its level of compatibility) as a guideline, 
> not as a strict dependency.  I can't tell if it's the strictness (which 
> has never been the intention) or something else that is bothering you.
> 
> The plan was to indicate that package A depends on package B and then to 
> list known working configurations, so: {{A.1.0, B.1.0}, {A.1.1, B.1.0}, 
> ...}.  If version 1.1.1 of A comes out and is marked as highly 
> compatible it would probably be installed automatically by default.  If 
> 1.2 came out and was marked as mostly compatible, it might prompt you. 
> If 2.0 came out and was marked as likely not compatible it would 
> probably not install it by default unless you forced it to.  But nothing 
> ever prevents you from installing the newest version you want.

Indeed. Nice to see that someone at least :) understand what I want and
how it would work.
IMHO the first step is collecting as much useful information as possible
and as easily as possible.
I have even toyed with the idea of autosubmitting successful SUnit
testruns.

Anyway, then we have the engine on top of this information and its
purpose is to *help* the user in deciding what releases should be
installed. This engine could very well be parameter controlled so that
for example I could tell it that I want it to be "paranoid stable" or
"bleeding edge, go for it" etc.

> I'm not saying it's necessarily perfect or easy, but I like it quite a 
> bit more than what you seem to be describing and I'd give it a fighting 
> chance of working.  But, like I said, I'm still not sure *exactly* what 
> the problem you have is, but I have a vague sense that you are missing 
> this part of what Goran has planned.  If I'm wrong and you already got 
> that part, then please forgive me :)
> 
> Julian

regards, Göran