Fwd: About the names of CPUs and architectures

[David Holmes]

On 18/06/2012 10:21 PM, Magnus Ihse Bursie wrote:
> On 2012-06-18 14:07, David Holmes wrote:
>> Magnus,
>>
>> You need to clarify exactly in what context you plan to make these
>> changes. For example, i386 is the "libarch" name for 32-bit x86 and it
>> probably has to remain the same for backward compatibility.
>
> Of course. I tried to be clear about this but appearantly didn't
> succeed. :)

To me the starting point here is to list all the different uses of these 
"architecture" names, as I started to (it may not be complete eg naming 
within makefile variables). Once you've worked out all the different 
uses, you'll have a better idea of how many different names are 
potentially needed and which ones have compatibility implications.

And of course ideally you only want there to be a single "input" 
variable name to define the architecture that is being built for.

David
-----

> Things that need to remain the same due to backwards compatibility needs
> to remain the same. I don't want to change that.
>
> My proposal was for everything that we can (easily) change, most
> typically in the newly written code in the new build-infra system, but
> potentially older stuff that don't affect backward compatibility if it's
> easy to fix (I don't count moving files around as "easy to fix"). I most
> of all wanted to set a standard for going forward.
>
> Second, I wanted to combine all the logic for "legacy" names in one
> place and make it clear; and hopefully get it as simple as possible,
> instead of having the "conversion" spread out across the system. My
> vision was to just list explicit exceptions to the (new) rule, like "for
> x86_64, the libarch is named amd64".
>
> Hope that clear things up!
>
> And thank you for your list of where the different platform names are
> used. I didn't know about all of them (like the i486.make file).
>
> /Magnus