On Thu, 2005-11-10 at 00:50 +0530, Arun Raghavan wrote:
> Call me an idealist, but this sounds too much like a "guilty until
> proven innocent" thing. I'm a "normal" Gentoo user, and had absolutely
> no trouble following the stage1 instructions. Let's leave the "the
> user is dumb" assumptions to Microsoft.
As I've said 1000 times now, you wasted your time if you followed the
Handbook. The whole point is that following the Handbook's instructions
for a stage1 installation gains you exactly nil, but costs you vast
amounts of time. The entire purpose of a stage1 tarball has been
distorted by countless people recommending it as an installation method
for everyone. The purpose of a stage1 tarball was for customizing your
bootstrap yourself. This means *editing* bootstrap.sh and making your
own changes to enact a build that wasn't envisioned by the Gentoo
developers. As you can guess, this isn't exactly something we had
documented, and for good reason. There are endless possibilities on
what you could do in this script. The wording in the current Handbook
on this really isn't strong enough. Were it up to me, it would say
something along the lines of:
"Stage1. Don't do it. Ever."
It would probably also be in big, red, flashing text.
I'll give you a good example. I have my own local bootstrap script that
causes my machines to build multiple cross-compilers during the
bootstrap process. It is this level of customization that cannot be
reached during this early phase of installation that the stage1 tarball
is meant to be used to accomplish. Running "emerge --sync &&
scripts/bootstrap.sh" is not empowering you to do anything but cause
possible problems which are then pushed onto the developers as "errors"
and "bugs", which they really are not.
> Instead, a prominent explamnation and warning about what doing
> so-and-so would entail would be good.
As I have also stated before, I am looking at completely removing the
stage1 tarball from distribution in the future. I simply no longer have
the time to troubleshoot the countless errors that are reported as bugs,
that are really configuration issues from the users themselves. Also,
the net gain of providing these tarballs is at best minimal, and at
worse very negative.
We are all volunteers here. Wasting what little time we have donated to
Gentoo tracking down problems with a broken process is not the best use
of our time. Making new features and improving quality is, though. Any
time that I spend working on a stage1 issue is time that I am *not*
spending testing and developing.
Release Engineering - Strategic Lead
x86 Architecture Team
Games - Developer