[Ecls-list] architecture-specific *FEATURES*

[Stelian Ionescu]

On Wed, 2009-11-11 at 08:34 +0100, Juan Jose Garcia-Ripoll wrote:
> On Wed, Nov 11, 2009 at 6:16 AM, Matthew Mondor
> <mm_lists at pulsar-zone.net> wrote:
> > Note that I've not looked into the implementation details yet to
> > suggest a diff, although I'd like to know if there's any existing
> > precedent about this I'm unaware of, or if the suggestion is welcome.
> 
> Even if you supply a CFLAGS, ECL still may need to guess the
> architecture it runs on when building, so the logic would not be
> "check CFLAGS and do that or this", but rather decide whether a
> platform / chipset feature is added at all at compilation time or at
> run time. The problem with the latter is that the logic for platform
> detection a-la-gnu is very very complicated: just have a look at
> config.guess / config.sub So in the case of run time detection via
> uname() it is not realistic to provide standardized platform names as
> features, but rather rely on whatever uname() gives us

I think that Matthew was only talking about detecting the CPU type; in
that case using the CPUID[1] instruction should be enough

[1] http://en.wikipedia.org/wiki/CPUID

-- 
Stelian Ionescu a.k.a. fe[nl]ix
Quidquid latine dictum sit, altum videtur.
http://common-lisp.net/project/iolib
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <https://mailman.common-lisp.net/pipermail/ecl-devel/attachments/20091111/d870d426/attachment.sig>