Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!ucsfcgl!cca.ucsf.edu!wet!epsilon
From: epsilon@wet.UUCP (Eric P. Scott)
Newsgroups: comp.sys.next
Subject: Re: Loading the /etc/services file into
Summary: nidump is buggy, and names can map to multiple services
Message-ID: <209@wet.UUCP>
Date: 26 Jun 89 02:21:05 GMT
References: <3539@cps3xx.UUCP> <800009@mrcnext.cso.uiuc.edu>
Reply-To: epsilon@wet.UUCP (Eric P. Scott)
Organization: Wetware Diversions, San Francisco
Lines: 29

In article <800009@mrcnext.cso.uiuc.edu> carlson@mrcnext.cso.uiuc.edu writes:
>Obviously, NetInfo can only store one entry for a key ("isis").

Wrong.  Look at it using the NetInfo application or niutil.
nidump is broken.

>Do you really need all 3 entries? Or just keep the tcp one?
>You might try to add suffixes to isis, like isis_tcp,
>although you will then have to patch the isis code.

RFC 1010 specifies that port number assignments should be
consistent across protocols; for example, "domain" is both 53/udp
and 53/tcp.  This means that  nidump services  should output
all port/protocol properties for a given key.  (Yes, isis can
have different port numbers for tcp and udp (but why?); RFC 1010
only speaks for ports <256.)

>This does seem like a real pain, I admit. 
>(Though so far we haven't noticed any problems.)
>Maybe the /services table should have a sub-table for each protocol,
>i.e. /services/tcp & /services/udp.

The further NeXT diverges from BSD compatibility, the more
attractive Sun (et al.) workstations become.  Hint, hint.
You better have a *good* reason for proposing this.

					Eric P. Scott
				San Francisco State University
				EPSCOTT@CALSTATE.BITNET