Re: [isabelle] Length of Proofs

You still have the capacity to code your own proof procedures; ML is just one level down. And you get a much more readable proof language.

Some years ago, I translated a lot of proofs from HOL to Isabelle. The HOL proofs were opaque blocks of tactics, while in the corresponding structure proof you could see (even if you knew little of the Isar language) which local facts were available for use and what had to be proved. And almost certainly, creating those opaque blocks of tactics required much more effort, because the old ML style (whether in HOL or Isabelle) isn't very good at forward reasoning.


On 21 Nov 2012, at 22:31, Jeremy Dawson <jeremy at> wrote:

> Given that in HOL (as in Isabelle, both pre and post Isar) one can always use proof procedures coded by someone else, this seems to sum up to saying that Isabelle/Isar is like HOL, minus the capacity to code one's own proof procedures, plus -- what ??

This archive was generated by a fusion of Pipermail (Mailman edition) and MHonArc.