To Jason Thorpe <thorpej@nas.nasa.gov>


>This is a perfect candidate for a COMPAT_11/COMPAT_10 compatibility 
>clause.  I don't see any good reason why the ability to run the old 
>programs should go away completely.

Please note that the binaries in question are actually either pre-1.1
binaries, or binaries produced by a ld which came from a pre-1.1
snapshot. I'm not sure what to call them. (COMPAT_10, I guess,
but there never _was_ a NetBSD/pmax 1.0 release...)

In any case, there really shouldn't be many of them around.
These binaries have other disadvantages, like (IIRC) that the GNU
binutils strip can't strip them.

I think I've convinced Chistos that a better long-term fix is to
generalize the MI code to be able to load what are perfectly
resepctable ELF binaries; but just not output the way a traditional
GNU binutils elf linker would produce them.


>Maybe a good name for the old ELF code would be mips/mips/elf_compat.c

True, if we decide to keep TEd's old implemetation.  In a panic to get
_something_ done before the final freeze, I didn't think very
clearly...