Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 10 Nov 2002 12:58:38 -0500
From:      Mike Barcroft <mike@FreeBSD.org>
To:        Garrett Wollman <wollman@lcs.mit.edu>
Cc:        standards@FreeBSD.org
Subject:   Re: Why I am opposed to a Standards Ghetto
Message-ID:  <20021110125838.A52940@espresso.q9media.com>
In-Reply-To: <200210262316.g9QNGAWB026174@khavrinen.lcs.mit.edu>; from wollman@lcs.mit.edu on Sat, Oct 26, 2002 at 07:16:10PM -0400
References:  <200210262316.g9QNGAWB026174@khavrinen.lcs.mit.edu>

next in thread | previous in thread | raw e-mail | index | archive | help
[Sorry for the late reply.]

Garrett Wollman <wollman@lcs.mit.edu> writes:
> What it boils down to is that we end up with two completely separate
> versions of every utility for which a standard exists -- one which
> implements the standard behavior, and one which does not -- for all
> time.  Broken scripts will never get updated, because script-writers
> will hard-code them to ignore the standard utilities and use the
> non-standard ones instead.  (We see that already in Solaris, where
> many users are totally unaware of the way in which their default
> environment differs from what the standard says, blithely trusting in
> Sun's conformance claim on the ticklist without investigating all the
> myriad actions one must take to get a compliant programming
> environment.)
> 
> That way lies madness.  Every question about a standard utility will
> then become unanswerable without a piece of information users would be
> surprised and in some cases hard-pressed to answer.  Ultimately, I
> think it's less of a POLA violation to tell our users, ``this used to
> work this way, but now there's a standard that says it works that way
> instead, and we wanted to comply with the standard'', than it is to
> encourage the creation of scripts which only work when one of multiple
> official FreeBSD versions of a utility is found in the search path
> first.

I agree that our aim should be for full compliance to the base
standard, but I don't think there's enough support from the rest of
the developer community to opt for the standard behaviour in the case
of some conflicts.

Taking this into consideration, I feel the next best thing would be to
provide alternative versions of utilities and make a POSIX version of
FreeBSD that could be built with a knob at `make release'-time.  (This
should shed some light on why I prefer /usr/posix/bin with seperate
manuals.)

> Note that I am specifically speaking of the base POSIX standard.  I am
> not including in this discussion the X/Open System Interfaces option
> of POSIX.1-2001, which is in essence the old System V Interface
> Definition.  To the extent we can support XSI behavior without
> conflicts, I think we should do so.  I do not think that we should
> supply a separate XSI version of utilities which do have significant
> conflicts (like `ps'), since our aim is not to emulate System V, and
> most of the differences as remain are old System V mistakes.

Yes, I think this is outside the scope of the current project, but
perhaps we might want to investigate this in the future.  I think it
would be of benefit to provide an alternative XSI (SYSV) version of
ps(1) for transitioning SYSV users to BSD.

Best regards,
Mike Barcroft

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-standards" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20021110125838.A52940>