TCP-group 1994
[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
Re: NOS is evil
- To: tcp-group@ucsd.edu
- Subject: Re: NOS is evil
- From: jmorriso@bogomips.ee.ubc.ca (John Paul Morrison)
- Date: Thu, 8 Dec 1994 15:35:33 -0800 (PST)
- In-reply-to: <6781.199412082229@lurpac.lancs.ac.uk> from "Mike Dent" at Dec 8, 94 10:29:43 pm
>
> > level process, I don't know how many 100kB it takes up to run) under
> > Linux! But Linux already *has* an excellent implentation of TCP/IP
> > and KERNEL level support for KISS tnc's and PI cards. For amateur
>
> Sure, but it has no real support for Joe Ham out there! At the moment
> as far as I am aware there are no packages that come close to matching
> Linux/JNOS for a user friendly front end, i.e. user logging into a
> BBS reading mail, sending it, gatewaying (netrom, telnet, AX25),
> and *heaps* more!
The point is that you don't need to run a process with its own tcp/ip
and ax25 stack, and internal multitasking, memory management etc. just
to run a BBS. The reason NOS has everything but the kitchen sink built
into it, is because of the limitations of DOS. Unix doesn't have these
limitations: it provides all the networking, filesystem, memory and
process management.
I don't understand the attraction of Internet BBSs, but there are a
number of internet bbs programs for Unix: you setup a userid of bbs,
with a login shell pointing to the bbs, and then the bbs goes from
there. I've also seen Lynx (the WWW client) patched up and used as a
login shell; this provides a nice user interface, ftp, telnet, WWW
etc. But at 1200bps this wouldn't be that fun.
I read that an AX.25 FBB or PMS style BBS was being worked on, it
runs as a single process (I imagine it forks for new connections).
This is how it 'should' be done.
> I run Linux/JNOS here and have done for several months now, it runs
> a treat and my users love it! I think they would be a little baffled
> if all they got was '$' or '%' !
> Mind you if you are willing to write something :-)
>
>
> > radio TCP/IP, it works great (for run of the mill AX.25 and BBS
> > support, it needs some polishing.)
> >
> > The mind boggles. I get a laugh when I see people using whatever hack
> > NOS has for SMTP, when they have the far more powerful Berkely
> > sendmail or Smail sitting under their noses. <Sigh> I guess old
> > habits die hard. Time to do some re-education...
> >
> Hmmm, sounds like you are severely knocking the NOS implementation of smtp
> here! DOes it not work for you?? I use JNOS smtp for my incoming
I don't know, I don't use it. Why would I when I can use Sendmail or
Smail?
> mail, and then use an alias to forward it to my Linux Sendmail. WOrks
> very well! Also you get the added benifit of LZW compression over
> the radio!
The XLZW extension isn't very useful: noone on the real internet uses
it; it's completely non-standard. I can turn the tables around, and
ask whether JNOS smtp supports MIME and ESMTP; I don't think it does.
(more people use ESMTP than will ever use XLZW)
> My outgoing mail, I use ELM and Sendmail. If it goes over the radio
> I let JNOS handle it. If I am sending to a non JNOS station I let
> Sendmail take care of it. Works fine for me!
It's good that it works, I just question the value of using it when
the builtin kernel TCP/IP and AX.25 provides all the low level
networking services you need. It's a shame to use Linux's TCP/IP just
for the loopback interface; and I think it's bizarre to be using
pseudo tty's configured in slip mode to route packets into a user
level process to duplicate services the kernel provides....
It must also be a pain to maintain JNOS sources; I mean, what if I
want to use Wuarchive ftpd? I'd have to hack JNOS to add it, recompile
and restart the whole damn thing. But I don't need to rebuild the
Linux kernel to support an ftp daemon, instead I just copy it to
/usr/sbin, edit inetd.conf and hup inetd, which makes more sense.
IMHO, I think (Linux) development work on NOS is a waste of effort:
either work on the kernel or work on porting clients and servers as
separate user level programs (but I'm not trying to tell people what
to do with their time.)
---------------------------------------------------------------------------
BogoMIPS Research Labs -- bogosity research & simulation -- VE7JPM --
jmorriso@bogomips.ee.ubc.ca ve7jpm@ve7jpm.ampr.org jmorriso@ve7ubc.ampr.org
---------------------------------------------------------------------------