Peeve: why "i386"?

[David Schultz]

On Thu, Jun 05, 2003, Rahul Siddharthan wrote:
> Why do all the BSDs continue to refer to the 32 bit Intel architecture
> as i386 even when they typically won't even install on an i386 any
> more?  Why not call it x86, or ia32, if not in the kernel config then
> at least in the release notes and documentation, as everyone else has
> been doing for years?
> 
> I personally find "i386" ugly and antiquated-sounding; many people
> find it confusing and misleading.  (Yes I know it's come up on the
> lists before.  I haven't seen any good answers though, "for historical
> reasons" isn't a good answer.)

The reasons for keeping with the i386 name have little to do with
tradition, as some people have implied.  i386 is the name of the
*architecture*.  The Intel 80386 was the first processor to
implement that architecture, and the latest Pentium 4 also
implements the architecture, albeit with a number of enhancements
over the previous generation.  The term IA-32 didn't come along
until a few years ago.  (1994 was when Intel first started work on
the design of the Itanium, and the marketing people didn't fiddle
around with the naming until a few years after that.)  So here is
a concise list of what I believe are the real reasons we don't use
something else:

	- ``i386'' is correct, as explained above.

	- Others use it too, including (I think) Solaris, which
	  doesn't support anything earlier than a Pentium.  IIRC,
	  the same is true of Linux.

	- Changing things now would be a major PITA, taking hours
	  of repo-surgery and scads of patches.  There's no good
	  reason to do this.  The fact that you personally find the
	  term ``Ugly and antiquated sounding'' certainly isn't a
	  justification, although I respect your opinion on the matter.

If people want the documentation to say ``x86'' or ``IA-32'',
that's another matter, but I would suggest that the documentation
remain consistent with the code insofar as there is the potential
for confusion.