Return to BSD News archive
Xref: sserve comp.unix.aix:48700 comp.unix.bsd:15538 comp.unix.pc-clone.32bit:7710 comp.unix.solaris:28165 comp.unix.unixware:15070
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msunews!agate!howland.reston.ans.net!swrinde!sgiblab!sgigate.sgi.com!fido.asd.sgi.com!slovax!lm
From: lm@slovax.engr.sgi.com (Larry McVoy)
Newsgroups: comp.unix.aix,comp.unix.bsd,comp.unix.pc-clone.32bit,comp.unix.solaris,comp.unix.unixware
Subject: Re: Unix for PC
Followup-To: comp.unix.aix,comp.unix.bsd,comp.unix.pc-clone.32bit,comp.unix.solaris,comp.unix.unixware
Date: 11 Dec 1994 01:30:51 GMT
Organization: Silicon Graphics Inc., Mountain View, CA
Lines: 74
Distribution: inet
Message-ID: <3cdkob$4t1@fido.asd.sgi.com>
References: <199411210319.TAA18133@nic.cerf.net> <D0E32G.3x8@news.cern.ch> <MICHAELV.94Dec10124723@MindBender.HeadCandy.com> <3cd8fo$ns5@fido.asd.sgi.com> <3cdef1$q87@pdq.coe.montana.edu>
Reply-To: lm@slovax.engr.sgi.com
NNTP-Posting-Host: slovax.engr.sgi.com
X-Newsreader: TIN [version 1.2 PL2]
I said:
I hate to burst your bubble, but I worked at Sun in the systems
group for a few years (and then in the server group). They had
*no* regression test other than the binaries that shipped with the
OS. Since 5.x, they use the POSIX test suites but those (were) are
pathetic and certainly don't cover everything.
Nate said:
IMHO, this is totally contradictory to your statement I've pulled down
for reference sake. The latter statement implies there is NO testing
done except to make sure the binaries 'run'.
And I'll try again. My last effort because this is turning into a
mindless pissing match which is why I don't want to deal with the BSD
crowd - every exchange I have turns into this sort of game. It's non
productive. My exchanges with the Linux crowd have been far more
focussed on "what should we do to make it better" not "I'm right and
you are wrong". Hint, hint.
If you read the quote above, I said "They had *no* regression test".
A regression test is a pretty formal sort of thing, it means that
you are promising a certain level of functionality and stability.
One example of an easy to do regression test would be to boot the new
kernel using the old binaries. I put POSIX into SunOS 4.1 and I did
this sort of test: I would boot the 4.1 kernel on a 4.0 vanilla
release. This was a test of the kernel interfaces and it did indeed
uncover problems.
Notice that this is quite different than compiling the binaries and
the kernel together and testing them together. Do you see how? It's
an important distinction.
Sun didn't do this sort of thing when I was there. Not as any formal
part of the release process that you seem to think is so wonderful.
: >The original point was that Linux is less tested than the major vendor's
: >releases. That may be true but the gap between Linux and a commercial
: >release is much smaller than you seem to think.
: As was pointed out in the Linux newsgroups, the 'it compiles' is
: sometimes all the testing some of the releases of Linux get, and some of
: the releases don't even get that far. This is (hopefully) far from what
: some of the commercial releases get as far as testing goes.
And I'm trying to tell you that you are wrong in your hope. For lots
of stuff "it compiles" is as far is it goes. Personally, I don't find
that to be a big crime. The big crime in my mind is that as bugs are
found and test programs are written to check for those bugs, those test
programs are not placed in a regression test harness. You could build
a pretty reasonable test suite over time that way.
: These are touted by
: many Linux activists as one of the great strengths about Linux. 'Bugs
: get fixed fast'.
That is a great strength. The GPL is another great strength.
: My point is that just because something is claimed to be fixed in a
: sub-release doesn't mean it *IS* fixed, or that it's a correct fix.
: With a commercial release you are at least given some guarantee of
: nominal usefulness and usability, but with the individual 'bug-fix'
: releases of Linux there is NO guarantee. That's why it's called
: beta-testing. :)
Boy, this is just naive. Go buy a copy of Solaris, install it on your
PC, and tell that it "just works". I'll bet you that Linux runs on
more hardware than Solaris by a long shot. And I'll bet if you plot
turn around time on bugs, Linux is outdoing Solaris by a factor of two
at least. Just a guess, do you have data to prove me wrong?
--
---
Larry McVoy (415) 390-1804 lm@sgi.com