Subject: Re: #ifdef CRAY in libexec/telnetd
To: None <firstname.lastname@example.org>
From: Brian Ginsbach <email@example.com>
Date: 07/16/2003 11:44:47
* firstname.lastname@example.org <email@example.com> [2003/07/13 23:57]:
> >I'd say it don't do it! Isn't NetBSD supposed to be "highly portable"?
> >Why can't this extend to commands/daemons? Personally I feel removing
> >this stuff only marginally improves readiblity. The code is somewhat
> >unreadable regardless IMHO.
> the question is the meaning of "portability".
"Portability" is portability. Slice and dice it other ways if you
> are we targetting to make NetBSD operating system as a whole to be
> easily portable to new CPU architecture? the answer is yes. it is
> stated in www.netbsd.org webpage.
> are we targetting to make applications in usr.bin to be portable across
> multiple platforms? (this is what you meant) if so, every usr.bin/foo
> has to have GNU configure script. therefore, i guess this is not the
> "portability" netbsd is targetting.
Why not? Good code (which I believe has always been one of the
underlying principles of NetBSD development) should be portable.
I disagree with every usr.bin/foo having a GNU configure script.
(Gross! Why on earth GNU configure? It certainly is not the be-all
end-all of portability.) Portable code makes for good code. What
you maybe able to get away with under one OS/compiler combination
may bite you on another. It can help point out sloppy code.
> for instance, usr.bin/ftp is NetBSD-only, and we have separate
> distribution "lukemftp" which adds portability. can't we take the same
> direction if we need to make telnet/telnetd portable? (i'm not sure if
> i'd volunteer)
This is one approach I suppose. I think that this case is a bit
different based on the history of theses applications.
Is there a publicly accessible CVS repository for lukemftp?
> >I think the only vendor that has even come close to "keeping up"
> >with telnetd is Cray (probably since Cray contributed its changes
> >back to Berkeley and they were incorporated into the 4.4 sources).
> could you expand on what you meant by "keeping up"?
I fairly certain that the telnet/telnetd in Solaris and IRIX are
still based on the 4.3BSD sources. They do not include the changes
that Dave Borman (once of Cray Research) contributed back to Berkeley
and were a part of the 4.4BSD releases. I also believe that this
(not keeping up) is why, although I can confirm, the #ifdef's for
other OSes were added.
Brian Ginsbach Cray Inc.