PR:

Based on private mail with Bill.Houle@SanDiegoCA.NCR.COM (Bill Houle),
where he said (quoting me)

On Jun 7, 11:52am, Brian Behlendorf wrote:
} >I can't see why you'd ever want $MACHINE to be variable since
} >all NCR boxes are generic SVR4. I don't see why you couldn't
} >just move the 'library' test to the list of i486-ncr-sysv4
} >above....
}
} Hmm, true.  Or at least move them next to each other so it's obvious.  I
} don't want to break something if there's a reason they're separate like that.
}
} >PS: Should I be anal-retentive and point out that most systems
} >these days are Pentium based rather than 486, or is 'i486' simply
} >a convention with no real CPU significance? 'intel' might be a
} >better (more generic) designation, and I'm sure this applies to
} >non-NCR systems as well.
}
} I don't think it has any significance in the code - what does "uname -m"
} result in on your systems?

You ready for a good laugh? On the machine I am on this moment,
it returns '3435'. On the machine I compiled Apache on Fri night,
it returned '5648'. NCR MP-RAS stupidly reports the *model number*
of the machine it is running on; ditto for `arch`. Thus, `uname -m`
is going to be all over the map depending on the class of box.

That is why I said all NCR tests -- including the "library" test --
should probably result in "i486" rather than the machine 'type'(sic).
Or, better yet, "intel" if there's no significance to 'i486' other
than as a generic x86 label.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@81506 13f79535-47bb-0310-9956-ffa450edef68