This pulled a couple minutes out of the build, but core is still the long pole in the tent, at 12m on my laptop.
-----Original Message----- From: Jeff Eastman [mailto:jeast...@narus.com] Sent: Friday, April 22, 2011 9:52 AM To: dev@mahout.apache.org Subject: RE: problems with util test cases One easy fix would be to use the sequential mode processing for the cluster dumper tests. Running in mapreduce mode the test takes 3 minutes on my laptop. With sequential mode it takes 2m10s, all but 22s doing SVD and DSVD computations (3 tests). With those 3 commented out it runs most quickly. I will commit that change now. -----Original Message----- From: Ted Dunning [mailto:ted.dunn...@gmail.com] Sent: Friday, April 22, 2011 9:16 AM To: dev@mahout.apache.org Subject: Re: problems with util test cases Yes. Makes huge sense to open a ticket. On Fri, Apr 22, 2011 at 9:12 AM, Christopher Jordan <cjor...@gilt.com>wrote: > Does it make sense to open a ticket to fix these test cases? I am likely > going to be poking at the cluster dumper next week and might be able to > help. > > On Apr 22, 2011, at 12:04 PM, Ted Dunning wrote: > > > I would love it. Especially since it doesn't check the results. > > > > On Fri, Apr 22, 2011 at 1:49 AM, Sean Owen <sro...@gmail.com> wrote: > > > >> It takes about five minutes to run too, which is maybe 1/3 of the whole > >> test > >> time. Would anyone be offended if we made this a non-test test that > didn't > >> run by default? > >> > >> On Thu, Apr 21, 2011 at 10:37 PM, Ted Dunning <ted.dunn...@gmail.com> > >> wrote: > >> > >>> No. That isn't the issue. Several of these tests run faster and when > >> run > >>> in isolation they don't seem to complete or fail. They stop executing, > >> but > >>> no succeeding tests run. > >>> > >>> > >> > >
