> The current discussion was quite useful to me.  Although discussion of the 
> solution in advance of implementation start with interested parties would 
> have been worth to do.

Yes I agree and apologies for that.  When I started it looked to me like a few 
dozen lines would do it, on top regular register filters.  But then the 
implementation grew and I was already "in too deep" to think of other
solutions.  (Add to that my slightly premature assumption that a special 
"global" constraint/new earlyclobber would be too specific to be acceptable).

-- 
Regards
 Robin

Reply via email to