Debian 0.91 installation report (iwj)
Well, I got my new disk, and here are the bugs/infelicities I found
when installing Debian 0.91. I'm still using my old setup because
0.91 doesn't yet come with a news system (I'll talk to Ian about this
During the mounting phase before installing the base, I responded to
the "which filesystem" prompt with /dev//dev/sdb4 by mistake.
I didn't notice, and typed the mount point, /root/var .
dinstall called mount with those arguments as given, which said:
mount: mount point /root/var does not exist
which was inaccurate, and then dinstall said
Cannot mount /dev//dev/sdb4 on /rootvar
which has a missing `/'.
Is the broadcast address not x.x.x.255, rather than x.x.x.0 ?
If so the example broadcast address in dinstall should be fixed.
It would be nice if dinstall was a bit cleverer about reusing
information you'd already given it (your network number etc.)
WARNING: No automatic LILO installation is done here
is cryptic for newcomers, and the point of it is unclear to experts.
The ftape driver causes all sorts of weird messages about interrupts
to come up when rebooting using the newly created floppy. They are
likely to confuse some people. It may have been mistakenly fiddling
with my ethercard, SCSI controller or 4-port card.
My system locked up because the default IRQ of the ethercard
conflicted with cua0. I temporarily disabled the net stuff by
rebooting into my old setup and mounting the new root partition.
This is not a Debian-specific problem, really - there should be a way
to set the IRQ of a programmable card before ifconfiging it. If there
is Debian should give you the opportunity to use it :-).
dpkg should not be case sensitive when asking for a category
abbrevation. It should also tell you when you ask for a category that
doesn't exist, rather than just redrawing the categories list screen.
In general, the package selection interface is fairly horrid.
When one quits out of dpkg after selecting the packages it's not
entirely clear what one is to do next. Perhaps this will be
alleviated by a "how to install Debian" document. Alternatively the
message when quitting could be made more verbose.
After installing X one has to make the X -> XF86_xxx link. If one has
only installed one X server it should have that as a default, or allow
the user to merely confirm it rather than asking them to type in the
I said "yes" to making the whatis file, and it said it would make one
in the background, but it didn't.
I then tried
dpkg --integrity man
Checking integrity (blah blah) done.
/var/adm/dpkg/inst/man.inst: No such file or directory
In /etc/inittab the text
# NOTE NOTE NOTE adjust this to your getty or you will not be
# able to login !!
looks like it came out of the original init package - it doesn't apply
When one switches to runlevel 6 to run xdm the gettys on VC's are
turned off. IMO this is a mistake; they should still be available.
Also, the gettys on tty5 and tty6 are only enabled in runlevel 4; they
should probably be enabled in 5 and 6 as well.
In /etc/gettydefs, the VC's should have echoe enabled by default.
The copy of the FAQ in /usr/doc is out of date.
After installing X one should be told where to get an Xconfig and how
to configure it (OK, pointed to an appropriate HOWTO or whatever) and
also told what to do to get xdm to come up by default (changing the
When fsck goes through the filesystems it should say which filesystem
it is about to check (preferably block device and mount point).
If one logs in via xdm the default erase and kill characters are wrong.
The Xsession file in /etc/X11/xdm is rather weird: it should certainly
look for .xsession in the user's home directory before looking for
xinitrc (if it looks for the latter at all).
One may want the facility to set system-wide X resources, in which
case Xsession should xrdb -load /etc/X11/Xresources if it exists, and
then xrdb -merge rather than -load the user's .Xresources if it finds