>
> Also sprach Robert M. Hyatt --- 12:25 PM 1/10/99 -0600
> >> Note also that the long list of errors that go with the "returned error 111"
> >> are a (non fatal) bug in the rpc layer.
> >>
> >> > > RPC: task of released request still queued
> >> > > RPC: (task is on xprt_pending)
> >> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >>
> >> That bit is a real bug
> >
> >yes... this has been around a good while AFAIK... I don't remember
> >when I first noticed it, but it was happening on 2.1.x where X was
> >less than 100... and is still happening on my 2.1's and the 2.2.0
> >pre-release I'm running on my quad xeon...
> >
> >I try to ignore that which doesn't crash me. :)
>
> I had this problem because in Slakware 3.5 , rc.inet2 does a
> `mount -a -t nfs` before loading portmapper.
> Upgrading from 2.0.36 to 2.2.0pre9 exposed this problem.
>
That is exactly what I was talking about Slackware. It is just one of
those things which annoy me very much. Last time when I checked, it
still used the BSD style init, which is very bad IMHO. Even FreeBSD is
trying to move away from it and use something close to SVR4 init. There
are some talks among FreeBSD to find something similar to our SysVinit
since they don't want to do straight copy of Linux :-).
BTW, I have been working on gcc/egcs, binutils, libc, gdb and knfsd.
There haven't been any slight interests in my work from Slackware. As
the result, I kept getting reports saying xxxx doesn't work on
Slackware. I had to tell them either to upgrade those packages
themselves or get a different Linux system all together. At the end,
I just gave up and tell anyone not to use Slackware.
Now the suggestions to Slackware if they are interested:
1. Use SysVinit.
2. Use glibc 2.
3. Use my egcs for Linux if there is one. I am preparing for egcs
1.1.1/Linux.
4. Use my binutils for Linux.
5. Use my gdb for Linux.
6. If it supports knfsd, use my knfsd package.
If they do all those, I will change my opinion on Slackware. I don't
think they will :-(.
H.J.
