Closed GoogleCodeExporter closed 9 years ago
That test case doesn't seem to be in the main tree. Please can you darcs push
it?
I couldn't reproduce this, but discovered that amusing things can happen if one
of the parameters to a programming problem is a proof that the propositional
simplifier can discover...
Original comment by adamgundry
on 3 Sep 2010 at 7:38
Thanks for pushing BugLetLambda.pig.
The simplification rule for |ALL s ABSURD| was just wrong. Moreover, using de
Bruijn indices in the propositional simplifer is dangerous: we must remember to
increment them when bringing a term under a binder. There may be other similar
issues remaining.
We also now retain more name advice, at the cost of losing knowledge that
certain functions are constant after simplification. Name advice in problem
simplification needs more work: BugLetLambda.pig now shows three hypotheses
losing their names.
Original comment by adamgundry
on 6 Sep 2010 at 9:50
The main patch fixing this issue is
http://www.e-pig.org/darcsweb/darcsweb?r=Pig09;a=commit;h=20100906094243-e29d1-a
c86262baaf8280118e38a41536766e3fc134a85.gz
I have now also changed problem simplification to retain name advice for
propositional hypotheses that simplify to a single proposition, in patch
http://www.e-pig.org/darcsweb/darcsweb?r=Pig09;a=commit;h=20100906100925-e29d1-a
731bbed967b5eea361e3d2334e207efe0b3093d.gz
Original comment by adamgundry
on 6 Sep 2010 at 10:14
Another related fix:
http://www.e-pig.org/darcsweb/darcsweb?r=Pig09;a=commit;h=20100909092151-e29d1-c
2056110662801cc172958bf0468bcb6960505ae.gz
Original comment by adamgundry
on 9 Sep 2010 at 9:36
Original issue reported on code.google.com by
pedag...@gmail.com
on 2 Sep 2010 at 1:22