I agree. Here is an idea:
You have the basic Mandrake-core-OS
You have the stable programs (fully tested) as first
choices
You have the mostly stable
You have the bleeding edge programs
In the installer, these are color-coded so if
stability is most important you can stay with the
tried and true, certain programs you might feel are
worth the risk... and then there are programs for the
truly adventurous.
The biggest thing this will do is allow for informed
decisions. The user will know what chances they are
taking. That in itself can make a huge difference in
perspective. I like to play with bleeding edge stuff
but I would want an office server and desktops to be
completely stable (with a few choice mostly stable
programs).
--- Gr�goire Colbert <[EMAIL PROTECTED]>
wrote:
> Don Head wrote:
> >> "All we are saying
> >> Is give min a chance"
> (...)
> > The "bleeding edge" issue I actually like when it
> comes to
> > a number of projects; MySQL, Apache, PHP,
> FreeCiv,
> > KDE/Konqueror, etc. Yes, "bleeding edge" in a
>
> These are projects which have proven to be viable
> and well tested
> (version >= 1.0). So there is little risk to destroy
> what used to work,
> as long as you don't ship beta versions (eg KDE) or
> remove patches that
> people like (eg Eazel theme).
>
> Where it hurts, it's that Mandrake ship programs
> which are not even beta,
> with version number near 0.0.x : even the
> *developer* would not expect
> the prog to work, so what the point in releasing
> that to the public? I
> started to use Linux when Mandrake was born. Hence,
> I've tried almost
> every Mandrake since 5.1 (~Mandrake 1.0). Since
> then, what has really
> evolved in Mandrake are:
> - large projects, like KDE, GIMP, etc., what you
> mentionned above.
> - Mandrake specific tools (much better than
> RedHat's...)
>
> On the other side, what is always driving Mandrake
> down are the tons of
> untested and undebugged applications that spoil the
> very needed
> impression of quality. That's why I reproach the
> quantity of packages in
> Mandrake, and require a Minimal Install and/or
> Mandrake-Core OS.
>
> In the ideal world, developers would not ship
> broken programs. But we
> cannot blame the developers : it is very difficult,
> even for the serious
> programmer (like me <laugh>), to test his program in
> every aspect,
> especially in event-driven applications. The less
> than ideal world we
> live in, has only one long-term alternative : to
> create a clean minimal
> system (Mandrake-Core, or whatever you would call
> it). You will have
> only thoroughly tested applications, and *zero*
> broken stuff. To sum it
> up : "What is not known to work, does not ship".
> Simple, and very
> attractive motto for investors.
>
> > distribution with a lot of packages is even
> worse, but
> > that, IMHO, doesn't really have anything to do
> with
> > installation architectures, goes against my
> personal
> > beliefs (I use Mandrake *because* of it's
> > "bleeding-edginess" (and it's open development
> model, which
> > I *REALLY* like)), and is a discussion that
> should be taken
> > to another thread.
>
> Ok, that's what I did. =)
>
> About Mandrake open development model, I agree that
> it suits particularly
> right to Free Software, as long as you do NOT assume
> that users are
> testers. A few days ago, someone at MandrakeSoft
> sent a message stating
> "I've released that, but not tested it". What the
> point in releasing
> stuff you haven't tested? Is this what is called
> Quality Assurance?
>
> Gr�goire, translator (that does not hurt anything).
>
> PS : anyone ever tested the Polish(ed) distribution?
> If so, would you
> mind mailing me privately with what you think about
> it? TIA.
>
>
>
>
>
>
>
=====
SI Reasoning
[EMAIL PROTECTED]
gnupg/pgp key id 035213BC
__________________________________________________
Do You Yahoo!?
Get email alerts & NEW webcam video instant messaging with Yahoo! Messenger
http://im.yahoo.com
