Re: [gentoo-dev] just a little feedback...

[Aron Griffis]

Hello Tibor,

As far as I can tell you haven't gotten many answers to these questions.
I'm new to Gentoo myself but I'll take a shot at some of them.

Tibor Rudas wrote:	[Tue Aug 07 2001, 05:38:42PM EDT]
>  - lately I am experiencing great difficulties in doing an install from the
> 1.0_rc5-r4 build iso image...
> It worked perfectly well on July 27th where I did the emerge rsync
> and the bootstrap. The emerge system followed on July 30th and the
> machine is up and running...
> During the last week however I tried to install an identical machine
> and failed multiple times (and I don't have super bombad racing!).
> At first even the new August-based bootsrap file failed because
> gettext-0.10.39 was required but it was not (yet) on ibiblio (this seems
> fixed today however...). I then bootstrapped from the last July-based
> file which worked but emerge system stopped with some error message
> (can't quite remember what it was right now something with shutils I think).
> Having done this two times I tried to Install from 'portage-20010730.tar.bz2'
> which I had lying around from some earlier experiments. This compiled to the
> end but left me in a system missing vital components like grub,
> start-stop-demon etc.... The emerge on grub worked but it kept searching
> for /boot/grub/stage1 and returned a "file not found" even after I put the
> file there.(?) But after the reboot I only ended up in a maintenance console
> due to the lacking sysVinit tools...

I don't know the answers to most of this (probably a lot gets
fixed/broken between bootstrap files).  However I think I know the
answer to the grub problem.

First, note that by default Gentoo doesn't mount the /boot filesystem,
as a precaution to keep it safe from filesystem corruption.  If you want
to to post-boot maintenance on it, you need to "mount /boot".

The next thing is that grub is searching _within_the_/boot_filesystem_
for /boot/grub/stage1.  That means that, relative to the root
filesystem, it actually needs to be installed in /boot/boot/grub/stage1.
I personally think this could be tidier, but it works.  :-)

> Today I again tried the classical approach which found an abrupt halt again
> in the bootstrapping process. gettex was found OK this time but ./configure
> complained about an unknown machine type "i685-pc-linux-gnu" ->
> there is a typo in the new make.conf in the PIII/Athlon section...
> but emerge system then stops complaining about missing makeinfo
> somewhere along the way...

I hope that somebody has fixed this.  I didn't have this problem with
rc5-r4 on Athlon.

> Am I doing something completely wrong or is portage in a state which does
> not allow a complete build-install right now.

I don't know the answer to this either.  Portage is definitely in
a continuing state of flux, but it's been usable for me.

>  - autofs:   in the startup script in the generation of the pidfile-name
> the sed-command 's/\//./' should be changed to 's/\//./g' otherwise
> mountpoints like /something/with/more/than/one/slash is causing problems
> since just autofs.something.pid is created but the script then looks for
> autofs.sometihng/with/more.... and fails to start.

This has not yet been fixed.  I believe needs to be fixed by somebody on
the Server Team?

> - xinetd: in the standard configuration xinetd just puts three lines
> to the syslog complaining about unknown "time-stream" and "time-dgram"
> and no more services - exiting, just to be restarted by supervise and
> the cycle starts again. Maybe xinetd shouldn't be started
> automatically since this currently just fills the syslog.

The problem here is that xinetd is compiled with ipv6 support, but your
kernel doesn't include ipv6.  There are two parts to solving this
problem:

(1) You need a new xinetd, which I haven't sent to gentoo-ebuild yet,
    but I will sometime today.  There are more problems in xinetd than
    just what you mentioned, which hopefully the new ebuild will fix.

(2) If you want to build with ipv6 enabled, you will need to add it to
    your USE variable.  This is an as-of-yet undocumented use option,
    but the new xinetd ebuild will pay attention to it.

> - mozilla: the ebuild builds fine and I can start mozilla from
> a Konsole in KDE and get some mozilla-messages (no direct error
> message) but then ... nothing. I can just stop the process with ctrl-c
> and switch to konqueror/operea/netscape...

My mozilla build doesn't work either.  I'm of the opinion that there
should be an alternate mozilla ebuild that just uses the prebuilt
binaries (which work!) from mozilla.org.  I may create this if the
Desktop Team doesn't get around to it first.  ;-)

> That were the most important things I noticed in these past 10 days.
> I am sorry if this is the wrong place to put this and if I'm rather vague
> on some of the details but I am currently at home and don't have all
> the details handy.

This is the right place to post these kinds of concerns.  Thanks for
being so detailed!

Aron