On Sat, 2002-12-21 at 08:58, Sven Vermeulen wrote: 
> On Sat, Dec 21, 2002 at 05:35:19PM +0100, M. Zuelsdorff wrote:
> > I am following the the discussion in the gentoo-dev group for more than a
> > year now. All I see is "a problem with this" and "a problem with that".
> > Some days ago, something even appeared to be "really fucked up". My
> > question: When do you expect Gentoo to become a final usable release?
> 
> It is very normal that you frequently see problems arise on the mailinglist
> but rarely successtories. This is because of human nature: we will swiftly
> ask/seek for help (and the Gentoo Mailinglists are a good place to ask for
> help) but rarely just mail to tell it works...
> 
> Think of it: if you emerge 10 tools, and 9 of them work flawlessly, you won't
> mail the mailinglist about those 9 tools do you? No, you're going to mail
> about the 10th tool that doesn't work.

You've just highlighted one of the biggest problems with Gentoo: manual
problem discovery and resolution.  When a package breaks, someone must
(1) manually discover it, (2) search mailing lists for Gentoo and the
application, (3) search the forums for Gentoo and the application, (4)
attempt reasonable diagnostics to insure the problem is not just with
their system, (5) if they are competent they might try to solve the
problem, and (6) share their problem with the community.

Currently, the latest gtkspell ebuild is broken for me.  I assume it
worked for the person who modified it last.  This hidden breakage
remains until manually discovered and shared.

Installing Java for Galeon for 1.4rc1, was a nightmare.  The user
documentation tells you are on your own if you use the latest sun-jdk. 
Yet, the latest is all that will work.  So users go on to try and fail
with sun-jdk-1.3.  The user docs should be authorative or have warnings
that they are not.

Checking the mailing lists is hard if the problem is not new.  Search
engines don't sort by most recent articles first.  Additionally, threads
break on month boundries, making following a thread not much fun.

BTW, checking the forums did revealed a solution for Java.

I've only mentioned a couple of the problems I had, but I ran into many
more that I haven't had time to properly investigate and report. :(

It seems to me, to make Gentoo a painless solution, supporting so much
customization, an automated problem detection system must be put in
place.  For example, Mozilla has tinderboxs that report when certain
builds break.  Additionally, some sort of automated user feedback system
would be good too.  This system might report the configuration for
compile failures/successes and package test results as well.

Unfortunately, the nature of source based installs (slow turn around
time) exacerbates user fustration and makes problem diagnostics and
resolution time consuming.

Hopefully some attention will be applied to this with the benefits of a
higher quality distribution and less wasted time for everyone.

Sticking with Gentoo :),

-Arthur

--
gentoo-dev <at> gentoo.org mailing list