Hi,

Am Montag, den 18.07.2016, 10:39 +0200 schrieb Lars Hupel:
> > > - What about sessions that grow in size over time?
> > 
> > That is indeed important, although we have just ignored it historically.
> 
> Right. But how would we take it into account? What registers as a spike
> in build time for a session could either be a performance regression in
> Isabelle or growing material.

Wouldn’t that be quite obvoius from the a quick glance at the changeset
in question?

I manually watch https://perf.haskell.org/ghc/ when I see a regression
I apply common sense to decide whether to notify the authors, or
whether it is genuine.

Sure, something fully automatic and reliable would be nice, but having
numbers and graphs is still useful, even if “graph goes up →
performance regression happened” does not hold unconditionally.

Greetings,
Joachim

-- 

Dr. rer. nat. Joachim Breitner
Wissenschaftlicher Mitarbeiter
http://pp.ipd.kit.edu/~breitner

Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
isabelle-dev mailing list
isabelle-...@in.tum.de
https://mailmanbroy.informatik.tu-muenchen.de/mailman/listinfo/isabelle-dev

Reply via email to