On 03/10/16 17:31, Lawrence Paulson wrote:
> I’d be happy with this, if it’s achievable.
>
> I’ve only tried the testboard once or twice; I find it too easy to do the
> wrong thing when you have a choice of targets to push to, and I have a
> powerful machine so I prefer to test there. But a mechanism of the sort you
> describe (provided it’s easy to use) would be ideal.
Larry, can you describe your own build/test process?
I have just used the new Jenkins testboard for the first time, and was a
bit disappointed by the very long runtimes: about 1.5h each for Isabelle
and AFP.
Mira set out to deliver Isabelle tests "In 10min!" (Bavarians need to
think of the famous Stoiber speech here).
I am doing manual Isabelle tests on my own machin in approx. 30min,
which is already a bit painful. 20min is OK, 10min would be delightful.
At some point I need to convene with David Matthews again, to see if and
how we manage another jump forward in general performance.
Makarius
_______________________________________________
isabelle-dev mailing list
isabelle-...@in.tum.de
https://mailmanbroy.informatik.tu-muenchen.de/mailman/listinfo/isabelle-dev
