Tollef Fog Heen wrote:
My gripe with it is the pimping of AMD. It's a different class than
i386, since there you only have one letter and not the whole name, and
historical errors should not prevent us from doing the right thing
today. (Else, all linux ports should be renamed to linux-$ARCH, which I
don't think they should (pre-multiarch, at least. ;))
My memory tells me that "a64" is the name of an ARM-derived
architecture. Don't even be tempted.
Currently, if I want to run the architecture, I _have_ to buy the
processor from AMD. Odd that we want to have the technical name
indicate what processor you need to acquire in order to get it to work.
The real reason why the current name for the arch is vendor specific is
because, if you recall, AMD invited Intel to sign up for a vendor
neutral name that they could both use ... and thereby avoid confusing
differentiation in the marketplace. Intel refused, which indicates to
me that Intel _wants_ us to use two different names for the respective
architectures. If we are going to take vendor preferences into account
in any way whatsoever, we should take _both_ vendors (the real one with
products and the virtual one that has only released marketing
literature) into account ...
I propose that politicizers of this argument take the _correct_ approach
to _their_ perceived problem:
1. Shut up whining until there is a non-AMD processor available
2. Read up the documentation on what Intel is going to do different (eg
execute bit)
3. Start a second porting project for whatever Intel's name is at
release time
4. When multiarch comes along, the two will be mostly interoperable anyway
Just another $0.02 ... this is getting expensive ...
Alex.
