Re: [patch/rfc] Obsolete the i960 target

[Andrew Cagney <ac131313@ges.redhat.com>]

> Hey Andrew,
> 
> I'm a little confused on how one deals with target/host-specific files
> when obsoleting a target or host.  Sometimes we prefix all lines of
> such a file with // OBSOLETE, sometimes we just leave those files
> alone (like this patch for the i960 target does).  I'm tempted to go
> with the latter (since it reduces the amount of work to do).  Is that OK?

Ah, I edited ``slightly'' the i960 patch and removed the diffs that just 
add the OBSOLETE prefix to every line of a file.  The actual files will 
be given the OBSOLETE treatment.

Adding the OBSOLETE prefix is very useful (I'd otherwize not be doing it 
:-).  It stops a search for to specific variables, macros or functions 
identifying references in obsolete code.  Something like:

#!/bin/sh
find *  -name 'ChangeLog*' -prune \
         -o -name '*.o' -prune \
         -o -name '*.a' -prune \
         -o -name '*~' -prune \
         -o -name '.#*' -prune \
         -o -name 'CVS' -prune \
         -o -name 'kernel' -prune \
         -o -name 'diff*' -prune \
         -o -name '*.orig' -prune \
         -o -name '*.rej' -prune \
         -o -name 'new-*' -prune \
         -o -name '%redact*' -prune \
         -o -type f \
         -print0 \
     | xargs -0 grep "$@" | grep -v OBSOLETE

works really well.

enjoy,
Andrew

PS: It's quicker if the buffer is put into fundamental-mode