From: Frank T Lofaro (fl0p+@andrew.cmu.edu)
Date: 10/14/92


From: fl0p+@andrew.cmu.edu (Frank T Lofaro)
Subject: Re: Why is TCP/IP treated like it isn't really part of Linux sometimes?
Date: 14 Oct 1992 05:36:09 GMT

Excerpts from comp.os.linux: 14-Oct-92 Re: Why is TCP/IP treated l..
Daniel AMP Carosone@mull (453)

>Besides, if you want to beta-test the networking code, you don't need
>it to be in a distribution kernel - unlike, for instance, the scsi
>driver patches where some people couldn't install linux on their scsi
>drives to test it.

        Umm, what if one does not have a modem, does have a net connection, and
doesn't want to keep around DOS tcp/ip software so one can ftp the
kernel? Okay, so its not typical, but neither is having a CD-ROM on a PC
typical either, but it should be supported, as it is in the default
kernel. I am just saying have the same apply to TCP/IP.

        There wouldn't be a problem if all the drivers were in the kernel. That
would be standard UN*X practice. As for beta test, the support code for
EFS (which is still alpha test) is in the kernel by default, but not
tcp/ip. (By the way, after having used the EFS for a couple of months, I
think it should be considered beta, not alpha. It doesn't appear to have
major bugs).

        Please no flame war. I am just trying to point out that the standard
UN*X practice (which the commerical unices, and 386BSD follow) of
including everything possibly needed to have a usable system is a good
idea. Departing from that, really, IMHO, isn't.