When you change something in your stock file, does your winols recognize the checksum
changes? Just to exclude that winols itself has got a problem.
If giac did not mix up things totally, at least the table starting around 81'FBXX/81FCXX
should look similar to your stock file and still contain some of the checksums from your stock.
I hardly can believe that giac rewrote the checksum algorithms in the code to protect against
further changes, but if you find heavy changes outside the parameter sections, this could be
the case.