[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: missing /dev/devices -- specifically /dev/rst0



 
> Last night, I needed to retrieve something from a tape backup.
> This caused me to discover that /dev/rst0 is still missing from
> debian 0.91.  Other devices are probably missing as well.  I
> tracked down the proper major device number (9), and added the
> device.  I'm still concerned that such missing devices will be
> a problem to those new to linux who will have a more difficult
> time tracking down the information needed to add them.

Just another point, I don't know if you've already discussed this,
but the audio devices are also absent...

Another minor problem I noticed in 0.91 (first time I've installed
Debian) was that the /etc/smail.conf uses an illegal/unrecognised
variable $osname in the smtp_banner line. The config file generated
by /usr/lib/smail/tools.linux/mkconfig is correct, and substitues
the literal "Linux". The minor symptom of the unknown variable is
that the banner is displayed as garbage when the smtp port is
connected to. Oh yeah, the smail mkconfig program likes to place the
new config file in /usr/lib/smail/config, rather than in the
"expected" /etc/smail.conf. The mkconfig program *also* adds a line
to rc.local to run smail in the background, and also appends to the
inetd.conf file, where both of these additions have already been
made. Looks like some hacking on the mkconfig script is warranted...

A rather more serious bug is with the (now very annoying) fdisk that
seems to be distributed with all Linux packages these days. Firstly,
it gives many redundant warnings, such as >1024 cylinders and the old
"partition has an odd number of sectors" (along with the "+" after the
partition entry) from the Minix days.

Another problem is that it still crashes given "incorrect" geometry.
Some controllers, such as the Seagate ST-0x and TMC-8xx SCSI ones do
not support the HDIO_REQ ioctl() very well, and this causes fdisk to
exit with a floating point error, usually when attempting to display
the partition table for the first time. Obviously, this is going to
be very confusing to the newbie...

One other minor fdisk gripe is that it still tells the user to reboot
after the partition table has been modified. This is *not* necessary
for any reason under Linux, it's just a defect of other brain-damaged
systems that's been perpertuated. This warning can (and should) be
removed.

Finally, just one general "policy" question here. On my system, I
like to have all files owned by bin.bin unless there is a reason
for them not to be (ie: /home dirs, suid programs etc). The only
(and to me important) advantage of this scheme is that it becomes
very easy to spot "abnormal" permissions in a long directory listing.
Has this been discussed before? Is there any chance we might see
something like this become "standard"?

All in all, first impressions are very good. Well done guys,
especially Ian.

PS: Sorry for such a long message...

-- 
pat -- empty space is wasted space.



Reply to: