linux-mips
[Top] [All Lists]

Re: A question about architecture and byte order with RPMs

To: Greg Chesson <greg@xtp.engr.sgi.com>
Subject: Re: A question about architecture and byte order with RPMs
From: "Ron G. Minnich" <rminnich@Sarnoff.COM>
Date: Thu, 4 Dec 1997 13:40:15 -0500 (EST)
Cc: Alex deVries <adevries@engsoc.carleton.ca>, ralf@uni-koblenz.de, SGI Linux <linux@cthulhu.engr.sgi.com>, rpm-list@redhat.com
In-reply-to: <9712040917.ZM8768@xtp.engr.sgi.com>
Sender: owner-linux@cthulhu.engr.sgi.com
> I've been asking the compiler people for a storage class modifier that could
> be added to type declarations.  The modifier would indicate whether the
> data object is stored in big-endian or little-endian format.
> The compiler would generate byte swizzles or not, depending on whether
> the native execution mode agrees with the indicated storage class.

or you could always look at the gokhale/minnich paper from 1993 (SEDMS) in
which we built a compiler that did just that, conforming to the IEEE
1596.5 standard (which I helped get going along with David James). It's 
not that hard.

We did a bit more than that, maybe too much more, but still ...

I still think it's a good idea :-)

ron


<Prev in Thread] Current Thread [Next in Thread>