Ian Piumarta <ian.piumarta at inria.fr> wrote:

> The value returned (in Unix) is very precisely defined.  'configure' deals
> in "canonical" host types that look like <cpu>-<vendor>-<os>.  These are
> usually calculated automatically by script called "config.guess" (which is
> part of autoconf).

It sounds like a reasonable answer given the capabilities of
autoconf/etc. It's a pity one can't get a reliable answer more like
'linux-gnu RedHat 7.1' but if we can't, we can't.

I'll edit the swiki page to add the extra attribute for the windowsystem
type.

"Andreas Raab" <andreas.raab at gmx.de> wrote:

> > Perhaps we should decide on which spec we really mean and meet it?
> 
> This seems hard. I always considered the "OS sub-type" to be something that
> needs to be interpreted depending on ones immediate needs.
I agree that there are many problems in trying to be too precise.
However, the current spec on the swiki claims we should be returning
something like 'win 95' or 'win NT' rather than just 'NT' etc. That part
would be trivial to 'fix' if indeed we conclude to obey the swiki spec.
Note that that documentation says specific OS type, not os sub-type. I
have no problem at all with changing that just so long as we agree. 

I think we have three options so far:-
a) ignore this and leave it as is
b) change Mac & win code to return 'Mac OS 8.60' and 'windows 95' -
obeying the swiki page
c) change Acorn to return '4.15' instead of 'RiscOS 4.15' and change
the swiki page.

Votes? Pronouncements ex cathedra?

tim
--
Tim Rowledge, tim at sumeru.stanford.edu, http://sumeru.stanford.edu/tim
Oxymorons: Taped live