Re: Reductions during Bison error handling

[Paul Hilfinger]

 > Given where you are, maybe you would have access to some
 > implementation of Corbett's Pascal Auditor?  Although we didn't
 > struggle very hard, we did try to use his error recovery scheme in
 > Bison, but it was a failure.  I could not find any hint of how the
 > stack should be massaged etc.  Having an original .y file would help
 > understanding how it was to be used.

The consensus here is that the software is essentially lost, and would
not be particularly useful in any case. Corbett concluded that
generating semantic information solely for the sake of error recovery
wasn't worth the considerable trouble, although one might be able to
use existing information that the compiler happened to have on hand. 

Paul