Re: Install stopper.

[jfm2@club-internet.fr]

> X-Authentication-Warning: unix2.sihope.com: emaddox owned process doing -bs
> X-UIDL: e6519dc84436e87e667fb134c6d83f27
> 
> On 5 Feb 1999 jfm2@club-internet.fr wrote:
> 
> > > > Screen: "comps Error"
> > > > Errors:
> > >   package XFree86-devel at line 1400 does not exist
> > >                                 1628
> > 
> > Is this the only one, or are they more?  I simply cannot download
> 
> If you kept or recall my first install report
> there were three packages listed, now only this one.
> 
> But, I don't understand about the downloading.
> I don't have that package. I don't need it or want it.
> So, why is "comps" even bothering with it?
> 

Ah! Now I understand, it was packages you didn't download.

XFree86-devel is listed in the comps file.  Each component should be
self contained ie list every package needed for the primary task but
also every thing needed by the primary packages be it libraries or
software needed for install scripts except for software contained in
the "Base" component because this is ever installed.

The media can be CD, NFS, a hard disk, or an FTP site but it needs to
contain the _complete_ set of packages for an install without warning
messages.   People willing to download only a subset of the packages
can use FTP install

People willing to do debugging of the distrib would not want to do an
FTP install every time so either they download everything to disk or
disregard messages about packages they didn't download.  And in
ulterior tries only download packages who have changed.

> You need to keep the Indy site updated since some newbies
> might want such a package. CS students, say.
> 

The web site or the distrib proper?  I don't have anyone for
maintaining the web site.

> But, most of us should not be expected to have a personal
> mirror of stuff we don't want. And for me that's
> source/development packages and non-English docs.
> 

Right.  But I don't have a clear solution right now for people who are
installing from a disk containing an uncomplete distrib.

> > > > ------------------
> > > > Phase 2, during reboot from boot floppy written in Phase 1.:
> > > 
> > > Forgot this before, but still a problem:
> > > The Red Hat/Indy boot/supp disks read in fast (30 seconds?),
> > > but the reboot disk (written during Phase 1)
> > > takes about 5 minutes to be read.
> > 
> > > The boot/supp disks and reboot disk should all read in
> > > just as fast, not some fast and the other slow (~5 min.)
> > > 
> > > This _must_ be fixed.
> > 
> > The rescue disk?  Don't know I never needed it and it is untouched
>       ^ You're right, I didn't even mention that, 'cause I didn't
> use it, so I don't know how fast it would have been read.
> I only know about the three I mentioned.
> 

Both the rescue and reboot disks use LILO and that means BIOS and BIOS
is slooooow.  Curiously I have found Pentiums who are slower at booting
from a floppy than mere 386s.

I don't know a cure except to tweakle with the BIOS settings and results are
not guaranteed.


> > > -----------------
> > > Also, please remove the LILO timeout.
> > > 
> > > It's very hard to hunt-n-peck "l i n u x S" or whatever, during reboot
> > > when the keyboard isn't yet configured (I use dvorak-r) and LILO then
> > > decides to take off without waiting for us to finish typing first.
> > 
> > But why were you rebooting single user?
> 
> Let's be careful here.
> 
> Instead of asking "But why were you ...",
> you need to ask,  "But why is LILO ..."
> 

Don't take offence I was asking what went wrong.

> You see, I'm not the one timing out prematurely, LILO is.
> So, you have to ask, "why is LILO", not, "why were you".
> 
> I think you'll find the answer to, "why is LILO",
> is that there is a "timeout" entry in /etc/lilo.conf.
> Remove that line, and you remove the bug.
> 

I thought azerty was worst keyboard in world for typing blind.

The problem is that there are boxes who need to reboot automatically
in case of a crash so we can't just remove the timeout.  AFAIK each
time you hit a key then the timer is reset (unless I did something
wrong, I will check) but in a dvorak keyboard it is easy to exceed the
5 seconds timeout who is the default so I will increase it to 30
seconds or 1 minute.  I will tell you when it is fixed.

> (A newbie can't be expected to know such things.
> They would just answer your question, "why were you",
> and pretty soon both of you are off chasing this tangent
> about what the newbie did 
> rather than fixing what LILO does.)
> 

If you were already in this list then you remember that I want to
replace LILO by the booter of defunct Linux Universe.  It allows you
to select your boot keyboard in a menu without rebooting, it shows you
a countdown of how much time remains to you before rebooting, you
don't need to run any command when you overwrite your kernel or
install a new one (no box left unbootable because you forgot to run
LILO).  And this is only part of what it can do.

> > > ---------------------
> > > 
> > > > Last message on screen I saw said something about "Appletalk".
> > > > 
> > > > Then screen scrambled with "Usage: route ..." junk at various places
> > > > all over the screen.
> > 
> > This is linuxconf who is called in the last phase of the boot to set
> > some networking things.  But it tries to identify the distrib in order
> > to know what mechanisms to use.  I will rebuild it first thing.
> 
> Thanks.
> 
> > > -----------------
> > > 
> > > I did reboot with "linux S" which got to "bash#" ok.
> 
> But, since you asked, it should be obvious now why I tried
> the single user reboot. A newbie would not have thought of it.
> 
> Newbies don't hunt bugs. They have too much work to do.
> And so do I.
> 

Please don't be offended.  Because at one time you mentionned a Debian
kernel along with scrambled chars I thought you were having
multicolored unreadable chars in your screen and that it was related
to VGA font support in the debian kernel

There are still bugs that is what it the distrib is still unannounced.


> > > -----------------------------
> > Until I upload a fixed linuxconf do a
> > 
> > ln -s Independence-release /etc/redhat-release (I dont remember if it
> > is Independence or independence
> 
> No.
> 
> I have enough frustrations just typing in all this stuff
> for all this email. It's nice to know there is your workaround
> for a really desperate fast typing expert.
> 
> I'm not an expert.
> I don't type very fast.
> I need stuff that works, not workarounds.
> Just like a newbie.
> 

Be conscious that the goal is making a cooperative project aiming for
a Linux better suited to people who have nobody babysitting them.  And
that at this time this is not a polished version for general use but a
buggy one for testers and that when unbuggy we will be still _far_ of
the final goal.

I apologize for all the problems you had.


> > > > -------------------------------------------------------------------------   
> > > > 
> > > > With Debian 2.0 booted I checked my Indy partition.
> > > > 
> > > > I happened to notice /var/lock had permissions of 775 and ownership of
> > > > root.00000014.
> > 
> > root.uucp.  Normal in redhat and clones.
> 
> I don't get it. uucp != 00000014 !
> 

Not in the /etc/group file of the Debain distrib but here is the entry
in Redhat's /etc/group

uucp::14:uucp


> Debian 2.0 has root.root and permissions of 777.
> 
> Furthermore, the only other time I've seen strange ownership
> entries like "00000014" is with some buggy versions of lynx
> that clobber ownership stuff wholesale, but usually both parts.
> 

In RedHat uucp applies its locks directly under /var/lock so it is
group uucp for uucp being able to write in the directory and PPP
knowing that the modem is locked by uucp.

OK this is not the smartest thing in the world but at this stage I
can't nor want separate from RH except when absolutely needed.
Manpower is too scarce for that.

-- 
			Jean Francois Martinez

Project Independence: Linux for the Masses
http://www.independence.seul.org