In private conversation, there has been some confusion concerning this issue (stemming from the users mailinglist originally).
Let me clarify, quoting the tutorial on code generation:
--
Code_Binary_Nat implements type nat using a binary rather than a linear
representation, which yields a considerable speedup for computations.
Pattern matching with 0 / Suc is eliminated by a preprocessor.
Code_Target_Nat implements type nat by integer and thus by
target-language built-in integers. Pattern matching with 0 / Suc is
eliminated by a preprocessor.
--
Both approaches currently share the same preprocessor which strives to
eliminate all pattern matching on nat in function arguments by combining
0/Suc-twins. It naturally fails if no twin can be found but pattern
matching on nat is still present. For Code_Binary_Nat, the alternative
preprocessor suggested by Brian could be used instead: the binary
presentation still allows pattern matching on 0 / Suc n to be replace by
pattern matching on 0 / nat_of_num, but no pattern matching with
successive Sucs.
Florian
--
PGP available:
http://home.informatik.tu-muenchen.de/haftmann/pgp/florian_haftmann_at_informatik_tu_muenchen_de
signature.asc
Description: OpenPGP digital signature
_______________________________________________ isabelle-dev mailing list isabelle-...@in.tum.de https://mailmanbroy.informatik.tu-muenchen.de/mailman/listinfo/isabelle-dev
