On 28/10/17 22:26, Makarius wrote:
> We are presently testing Poly/ML 5.7.1 by default (see
> Isabelle/aefaaef29c58) and there are already interesting performance
> figures, e.g. see:
The daily "AFP slow" timing has arrived just now, 4h hours later than
with Poly/ML 5.6:
I still need to investigate, why some sessions require much longer now.
It might be due massive amounts of generated code.
isabelle-dev mailing list