From: dumas patrice (dumas@rip.ens-cachan.fr)
Date: Thu Mar 30 2000 - 12:18:00 CEST
On Wed, 29 Mar 2000, Michele Andreoli wrote:
> On Wed, Mar 29, 2000 at 02:06:47PM +0200, dumas patrice nicely wrote:
> >
> >
> > Then I have some remarks :
> > It seems that there are errors in the Modules documentation : path are
> > said to be in /etc/modules.conf, but they are in /etc/modules.path.
>
> This is changed a little from various release. Do not trust totally
> in readme.html, but check always HELP in muLinux itself.
This is from the help (in /usr/doc/...)
> The /tmp/modules works as "cache space" for modules, very important
> for RAM muLinux: a modules is unpacked ONE times and left in this
> directory.
>
> /etc/modules.conf (previous name) conflict with a file required my
> "modprobe", then I changed name.
>
> > it says to put the object file in the DIR specified in
> > /etc/modules.path, but I think it's not true, it should be in
> > DIR/preferred. Note that I could be missing something here.
>
> Yes. A documentation error. Actually, Setup search for DIS/preferred
> but preferred is a link to 2.0.36. You must create 2.0.36 directory
> and put your module in it (this is for *unpacked* module). Anycase,
> it is better to put your modules in archive.tbz.
I wanted to do this, and so use unpack then pack, but in the
documentation, it was said that it will crash with the ramdisk
installation. In this case, is there a way to incorporate my module in
archive.tbz and have it installed in my setup on floppy disk ? (I should
read the code, in fact, so if you want to let me answer my question
myself, it's possible...).
> > I also have a little critic, that is I think the message that appears at
> > the end of runLevel 4 is a bit confusing, because level 5 isn't only about
> > configuring but also loading the existing configuration. So the second
> > time I boot, I stopped at this message, thinking my stored configuration
> > had been totally loaded, but there wasn't the things configured in custom,
> > and I didn't understood why.
>
> The message
>
>
> ==================================================================
> ** Press -ENTER- in next 5 seconds to stop autoconfiguration **
> ==================================================================
>
> is emitted at the top of runlevel-5. If you stop, system do not load
> configuration. Following behaviour depend from kind of muLinux you
> run:
>
> ram-muLinux) because no profile is loaded, the system run with FACTORY
> defaults, i.e. what I cabled in /setup/cnf for my convenience.
>
> hard-muLinux) illogical behaviour; the system use the last saved /setup/cnf,
> because this file are permanent.
>
> I provided the stop, in the case the of blocking profile (name='lock') and
> in the case this profile is corrupted or hang the system. Using the
> stop you can fix the problem.
>
>
> On the other hand, the exact point where to put the stop is matter
> of discussion, and I haven't solution. If I put it *after* profile loading,
> but *first* of activate it, what means really?
Well I think you didn't understood my concern (I am not an english
natural tongue speaker too, as I am french, and I am not always clear !),
I understand perfectly why you ask for a confirmation before going through
the runlevel 5, but the critic was about the message itself. I will
provide an example of what (in my strict opinion) should be better :
******************************************************************
Press enter in 5 seconds to stop here the configuration and the loading
of your allready configured custom setup.
***************************************************************
It is true that the other message is smaller and has the same meaning
because autoconfiguration means the two things, and so it's just a
proposal, as I was confused last time.
An other thing is that I think running the syst test before the end of the
configuration is a little risked, as there is a chance that it crashes,
before the setup is saved, and before you can switch to another console or
^C it. It happend to me with the test of the printer, it is an ash bug I
think as it is the result of something like "..." >/dev/lp0 .
To fix this problem, I think there are two possibilities, one being to put
an add to the message when it asks whether you want to test the
hardware, like :
"message" (risky, you can lose all the work you did before)
The other to launch it in a separate script, during the tour made at the
end of the install.
Oh, I forget, where can I get the source of, for example muless, or other
nice little programs of mulinux ?
Pat
---------------------------------------------------------------------
To unsubscribe, e-mail: mulinux-unsubscribe@sunsite.auc.dk
For additional commands, e-mail: mulinux-help@sunsite.auc.dk
This archive was generated by hypermail 2.1.6 : Sat Feb 08 2003 - 15:27:13 CET