On Fri, Mar 28, 2025 at 04:52:04PM +0100, Ingo Schwarze wrote:
> Hello Jonathan,
> 
> Jonathan Gray wrote on Sat, Mar 29, 2025 at 01:56:29AM +1100:
> > On Fri, Mar 28, 2025 at 02:55:31PM +0100, Ingo Schwarze wrote:
> >> Andrew Daugherity wrote on Thu, Mar 27, 2025 at 06:18:43PM -0500:
> 
> >>> Because I was curious, and Brian Conway had replied that this dated
> >>> back to the very first commit of OpenBSD in Oct. 1995
> >>> https://github.com/openbsd/src/blob/df930be708d50e9715f173caa26ffe1b7599b157/etc/netstart#L29
> >>> when it was > forked from NetBSD, I went looking through the NetBSD 
> >>> history.
> >>> Evidently support for /etc/my{name,gate} and /etc/hostname.* was added
> >>> in Apr. 1993, just before the release of NetBSD 0.8
> >>> https://github.com/NetBSD/src/commit/80e5f01ba42eb7e7bc57b9dfe71a889ab11f995c
> 
> >> Continuing the timeline backwards:
> >> 
> >> July 19, 1993 FreeBSD: Rodney W. Grimes adds myname(5) and mygate(5) 
> >> support
> >>               commit message:
> >>               "From NetBSD, copied verbatium.  May need some work yet."
> >>               https://github.com/freebsd/freebsd-src/commit/0a71fe69
> >>               This tells us that none of this came from FreeBSD.
> >> Apr 20, 1993  NetBSD 0.8 release
> >> Apr 02, 1993  cgd@ commits myname(5) and mygate(5) support to NetBSD
> >> Jul 14, 1992  386BSD 0.1 release
> >>               There, /etc/netstart uses /etc/myname (but not mygate)
> >>               https://github.com/386bsd/386bsd/blob/0.1/etc/netstart
> >>               Appears to be based on SCCS revision 5.9, but with 
> >> modifications
> 
> > no /etc/myname in the 386BSD 0.1 sources that were on minnie.tuhs.org
> > https://github.com/jonathangray/386bsd/blob/master/othersrc/etc/netstart
> > 
> > extracting the 0.1 etc files from another source also doesn't have it
> 
> Very strange.  In a copy of the 386.0.1 sources that i have,
> the file etc/netstart does contain
> 
>   hostname=`cat /etc/myname`
>   hostname $hostname
> 
> and all the meta files on the top level (like BUGNFIX.FORM, CAREWARE.INFO,
> CONTRIB.LIST, COPYRIGHT, DISTRIBUTION.HOSTS, INSTALL.NOTES, ROADMAP
> and so on clearly say "386BSD RELEASE 0.1" and "July 1992".
> 
> As far as i remeber, i did get that file from minnie.tuhs.org
> on February 21, 2014.  Given that your observations differ,
> either i made a mistake back then (but i don't quite see what kind
> of a mistake that might have been), or different versions of 386BSD 0.1
> existed.

It turns out netstart from the 386BSD 0.1 bin01* sets has the
change, but netstart from the etc01* does not.

0.1 bin:
# my-name is my symbolic name
# my-netmask is specified in /etc/networks
#
hostname=`cat /etc/myname`
hostname $hostname

0.1 etc:
# myname is my symbolic name
# my-netmask is specified in /etc/networks
#
hostname=myname.my.domain
hostname $hostname

0.1 src:
no netstart file

0.0 bin:
# my-name is my symbolic name
# my-netmask is specified in /etc/networks
#
hostname=my-name
hostname $hostname

0.0 src:
# myname is my symbolic name
# my-netmask is specified in /etc/networks
#
hostname=myname.my.domain

Net/2:
# myname is my symbolic name
# my-netmask is specified in /etc/networks
#
hostname=myname.my.domain

> 
> > If it were really in 0.1, there wouldn't be commits to add it to
> > NetBSD and FreeBSD.
> 
> That argument has some merit because the development that led to both
> NetBSD 0.8 and FreeBSD 1.0 was indeed based on 386BSD 0.1.
> 
> So i guess the most reasonable option is to just refer to NetBSD 0.8
> for now.
> 
> >> March   1992  386BSD 0.0 release; it uses neither; SCCS revision 5.9
> >>               
> >> https://github.com/dspinellis/unix-history-repo/blob/386BSD-0.0-Snapshot-Development/usr/src/etc/netstart
> >>               So 0.0 is apparently based on Net/2 (5.9) without 
> >> modifications.
> >> Aug 20, 1991  4.3BSD Net/2 release; it uses neither; SCCS revision 5.9
> >> June    1990  4.3BSD-Reno release; it uses neither; SCCS revision 5.9
> 
> > Why should people know the release these manual pages appeared?
> 
> I agree such information usually isn't needed.  In this case, i didn't
> remove it because (1) the point of the patch was adding missing
> information and removing something would only be loosely related
> and (2) the delay between introducing the feature and documenting it
> was unusually long.  Not very strong arguments, though.
> 
> So maybe like this?
>   Ingo
> 

ok jsg@ for this with myname changed from .Nx 0.8 to 386BSD 0.1

sorry for the confusion

> 
> Index: myname.5
> ===================================================================
> RCS file: /cvs/src/share/man/man5/myname.5,v
> diff -u -r1.12 myname.5
> --- myname.5  1 Jul 2021 01:16:13 -0000       1.12
> +++ myname.5  28 Mar 2025 15:25:27 -0000
> @@ -64,5 +64,7 @@
>  .Xr hostname 7 ,
>  .Xr rc 8
>  .Sh HISTORY
> -This manual page first appeared in
> -.Ox 3.4 .
> +Support for the file
> +.Pa /etc/myname
> +first appeared in
> +.Nx 0.8 .
> Index: mygate.5
> ===================================================================
> RCS file: /cvs/src/share/man/man5/mygate.5,v
> diff -u -r1.2 mygate.5
> --- mygate.5  17 Oct 2021 13:20:46 -0000      1.2
> +++ mygate.5  28 Mar 2025 15:25:27 -0000
> @@ -82,5 +82,7 @@
>  .Xr netstart 8 ,
>  .Xr route 8
>  .Sh HISTORY
> -This manual page first appeared in
> -.Ox 3.4 .
> +Support for the file
> +.Pa /etc/mygate
> +first appeared in
> +.Nx 0.8 .
> 
> 

Reply via email to