Comments with -fpreprocessed

[dewar@gnat.com]

<<but my point of view is that it *did* work before - it correctly rejected
an ill-formed program. 'accepting something that you did not before' is
a regression, if `something' is ill-formed.
>>

But something that was rejected before was NOT something that was working!
By your criterion adding a new upwards compatible feature would be a 
regression, since a previous attempt to use this feature would have been
rejected and is not rejected any longer.

Like Humpty-Dumpty in Alice, you can make words mean anything you want, but
you do not clarify your argument by using the word regression in a sense
that is so far out of the ordinary.