My concern is this - are users of bison ready to wade through y.tab.c
file (or in case of bison the xyz.tab.c) ?
I think that's a misunderstanding of what I proposed. My idea is that
the function for each state could have comments specifically designed
to be convenient for debugging.
The commercial tools available today (there aren't many by the way)
take an alternative route - they let you compose your grammar in a
nice IDE and then point out all the conflicts, simulate a parse on a
test input file even before the final parser code is generated.
That does seem like a useful feature. It is quite different from the
sort of debugger that I thought we were talking about before.
