[PATCH] Don't use slowcompare method unconditionally

[Jakub Jelinek]

On Wed, Nov 09, 2005 at 12:11:11PM -0500, Andrew Pinski wrote:
> Becuase there are much older machines which don't support "tail -n".

If you read the patch I posted, you'll notice that it can't break much
older machines that don't support tail -c+16 syntax, as it keeps using
tail +16c, just uses cmp --ignore-initial=16 or other method preferrably.
Even if tail supports +16c syntax, my patch still saves creation of ~ 330
unnecessary temporary files.

I have nothing against coreutils changing to support that syntax by default,
but I don't think it is a good idea for GCC to rely on it.  That just means
obstackles for GCC users.

	Jakub