Unlimited software warranties

[Ronald S H Khoo]

sef at kithrup.COM (Sean Eric Fagan) writes:

> peter at ficc.ferranti.com (Peter da Silva) writes:
> >Show me where it's being sold. How do I vote with my pocketbook if there
> >are no names on the ballot?
> 
> SCO XENIX.  Everyone wanted to go to SysVr3.2; SCO didn't create the market,
> you know.  Despite XENIX being very stable, small, and fast, people wanted
> "real" UNIX.  Because of the features, partially, and partially because of
> the name.

But I think that SCO XENIX gives evidence that in this particular market,
the "Classic Coke" formula doesn't work 100%, *but* there is a definite call for
it.  The problem is that computer hardware doesn't stand still.  For example,
SCO Xenix 2.3.2GT (the only version actually in production and on sale from
them) doesn't boot on some IDE machines, so they had to engineer a fix (xnx259).

There are other older examples, like the fact that 2.2.1 falsely detected
the existence of FPUs in some machines that were produced *after* the software
was released.  This made running awk (and hence /etc/custom) slightly
difficult, as that release did not have the ignorefpu kernel switch.
``fixed in the next release''.  What else can you do ?  You can't stop
hardware people adding features and making changes any more than you can
stop software people producing patches to their code.  And in the case of
hardware, you can't tell them to ship the equivalent of "copies of last
year's distribution floppies" :-)

On the other hand, people *DO* want Classic coke, which is why SCO is now
coming up with 2.3.4 of their Xenix a year after the product was killed :-)
If ISC take up Peter's challenge of providing Unix Classic, they will have
to realise that it will take more than zero development work, to get
hardware workarounds out, etc.  And they should definitely ship it with
the u-area bug fixed :-).  But I would agree that it's a worth while
thing to do, and it should be aimed at the budget market, i.e. cheap.
IE: "watch my mips, no new features" as it were, but hardware support
work must go on.  Sigh.  I suppose that means that the $150 unix doesn't
quite make it :-(

True "Classic Coke" would require a hardware freeze as well as a software one.
It is something that would make my employer's life much easier, for one.
Maybe it's worth someone's while to license redistribution of some Unix
or other at a stable release that works with their hardware, and promise
to ship that hardware for ever with that version of Unix ?  That would
work, assuming that the economics would work out.  That question I have
no answer to.  Maybe someone should commission a market survey and
do the sums.  But that's not a comp.* issue :-)
-- 
Ronald Khoo <ronald at robobar.co.uk> +44 81 991 1142 (O) +44 71 229 7741 (H)