Yes, we're now happy that it was this bug that our test hit, and that it's fixed in 3.1.1. +1 for a release then! Matthew Webber
> -----Original Message----- > From: Gilles Sadowski [mailto:gil...@harfang.homelinux.org] > Sent: 11 January 2013 09:32 > To: dev@commons.apache.org > Subject: Re: [VOTE][RC1] Release Commons Math 3.1.1 > > On Fri, Jan 11, 2013 at 09:10:43AM +0000, matthew.web...@diamond.ac.uk wrote: > > The vote is closed, I think, > > It's not yet closed; you are welcome to vote. > > > but the following observation might be useful: > > > > We were previously running with an earlier snapshot of 3.1.0 (from > > 2012/10/09). When we upgraded to 3.1.0 final, the runtime of one of our > > JUnit tests jumped from consistently 12 SECONDS to a *very* long time (I > > killed it after 30 MINUTES). We didn’t seem to be converging as quickly > > as we were previously ;-) > > > > The developer involved said that there were some issues with the test, but > > never-the-less it was clear that something had changed, and not for the > > better ... > > > > I re-ran the test with 3.1.1 RC, and the runtime is now consistently 27 > > seconds. > > > > This would probably be more useful if I could tell you what our test code > > did, but anyhow ... > > This sounds like the bug which this point release was aimed to fix (needless > creation of a dense matrix). If that was the case in your test, maybe that > implied swapping to disk. > > Regards, > Gilles >