> The revisions you site are simple fixed-size type fixups.  They aren't on
> the scale of hackery required to work around the -fshort-enums braindamage.

Oh come on.  Grow up.  -fshort-enums is a perfectly legitimate 
implementation of enums, just like using 64-bits for a long is perfectly 
legitimate.  If code isn't written portably then it's going to have to be 
fixed when it is ported.  That's a fact of life.  You can't define the 
problem away by just pretending it isn't there.

R.

PS. I've a better idea, lets make ia-64, alpha and sparcv9 use 32-bit 
longs.  After all, everyone knows a long is just 32 bits...