Re: cvs commit: fptools/happy/src LALR.lhs ProduceGLRCode.lhs

I forgot to mention:

I've also simplified the counting of clashes, with Simon M's agreement.

 * We count one S/R clash if the chosen shift action clashes with x >= 1
   reductions

 * We count one R/R clash if the chosen reduction clashes with x >= 1
   reductions.

 * Hence we don't count R/R clashes if there is already an S/R clash in
   that table entry; and we don't count how many reductions were clashing.

This is a simplification of the previous code, given the invariant
enforced by my recent change. Note that the purpose of counting is to
count existence of problems, and the programmer should consult the .info
file if they want more detail.

Paul