In the past few days AFP has been continuously broken: presently
AFP/3fc9e6ace21f and Isabelle/1f27d75ccf05. After some time in a bad state
it becomes increasingly difficult to guess which changes are responsible.
http://isabelle.in.tum.de/reports/Isabelle/shortlog provides some clues,
but they are unsure: it looks like lifting, codegen, potentially
codatatype.
Some years ago we used to have broken AFP states routinely, in
correspondance to full-day test time. Now that is around 1h, so there is
hardly a reason of not testing frequently, especially in conjunction with
pushes on main Isabelle. Time is better invested to test directly
(isabelle build -a -d '$AFP') instead of making remote guesses.
Makarius
_______________________________________________
isabelle-dev mailing list
isabelle-...@in.tum.de
https://mailmanbroy.informatik.tu-muenchen.de/mailman/listinfo/isabelle-dev
