Open MSoegtropIMC opened 1 year ago
@andrew-appel : sorry there is one issue with this PR - I did it on Apple silicon (ARM) - the symbol tables will be different for Intel and ARM. I can check this tomorrow - today I don't have access to an Intel machine.
I also need to think about a way to abstract this ...
FYI: I have to delay this a bit - I am working on the Coq Platform release (after the 8.16.1 release).
This PR adds the following functionality:
an Elpi based set of functions to create lists of symbols used for delta reductions with cbv
a set of Ltac2 functions and tactics which perform restricted cbv reductions
a similar set of reduction functions written in Elpi is also supplied, but this is experimental
the simplification by cbv is prepared for use in one place in file
floyd/go_lower.v
by replacing a call to simpl with a call to a dedicated tactic. But currently this tactic is still defined to usesimpl
because switching to cbv leads to some differences in goals left by the entailer.The point of this PR is two fold:
I expect to do another PR soon which contains a reflexive tactic for Z/Int range proofs - a common goal in entailment. Larger rework in the entailer caused by the move to cbv should wait until this PR.
Creating two new top level folders "elpi" and "ltac2" might be excessive - maybe these should go under floyd. I expect that long term VST will use more ltac2 and elpi tactics, so that having such folders makes sense. The Ltac2 standard library is a bit terse, so that it is common to have a bit of Ltac2 utility code, which needs to be organised.