On Fri, Oct 8, 2010 at 3:06 AM, Niall Pemberton <niall.pember...@gmail.com> wrote: > On Thu, Oct 7, 2010 at 12:21 PM, Niall Pemberton > <niall.pember...@gmail.com> wrote: >> On Thu, Oct 7, 2010 at 10:31 AM, Jörg Schaible <joerg.schai...@gmx.de> wrote: >>> Hi Niall, >>> >>> Niall Pemberton wrote: >>> >>>> I have prepared Commons IO 2.0 RC2 for review (rc1 never went past the >>>> tag). As there have been quite a few changes in the last week, I'll >>>> leave it a few days before even considering whether to call a vote, to >>>> give time for feedback. >>>> >>>> The distro is here: >>>> http://people.apache.org/~niallp/io-2.0-rc2/ >>>> >>>> Release Notes: >>>> http://people.apache.org/~niallp/io-2.0-rc2/RELEASE-NOTES.txt >>>> >>>> Site: >>>> http://people.apache.org/~niallp/io-2.0-rc2/site/ >>>> >>>> Maven Stuff: >>>> http://people.apache.org/~niallp/io-2.0-rc2/maven/ >>>> >>>> Some Notes: >>>> >>>> * There is one error on the clirr report - which is a false positive >>>> (a generic method that is erased) >>>> http://people.apache.org/~niallp/io-2.0-rc2/site/clirr-report.html >>>> * Links to the JavaDoc versions on the site don't work (they will when >>>> its deployed to the right location) >>> >>> Do you know why Gump is failing for "commons-io-test" since some weeks? >> >> There has been more than one reason for the failures :( >> >> Some times it was the FileSystemMonitorTestCase - which was also >> failing in Continuum about every 3rd or 4th build. I made some changes >> to that test and since then out of 43 builds it has only failed once. >> >> The other main cause is FileCleaningTrackerTestCase throwing an >> OutOfMemoryError. This was failing in Continuum and I made some >> changes which seemed to have reduced the failures but there is still >> the odd instance of it. Gump seems to be failing mostly because of >> this now (although it is passing some of the time). This test tries to >> fill up memory so that PhantomReferences get released - so thats >> causing the OutOfMemoryError and is a problem with the test rather >> than FileCleaningTracker >> >> I have updated the following JIRA ticket with the history of Continuum >> builds and the causes of the failures: >> https://issues.apache.org/jira/browse/IO-196 >> >> In summary though most failures are FileCleaningTrackerTestCase and >> the problem is the test case. > > I've done some additional work on both the FileSystemMonitorTestCase > and FileCleaningTrackerTestCase which I hope will get rid of the > errors. Gump has been pretty much daily, so should have an indication > soon. I have added the gump failures to the JIRA issue (from the mail > archive) - can't tell the cause as theres no history AFAIK with gump. > https://issues.apache.org/jira/browse/IO-196
Continuum has now run 30 builds without failing. Since the failures are intermittent, theres no guarantee that they won't fail again, but I believe that with the changes I've made they will at least happen much less. I put together the following page which shows the history of Continuum builds, which test caused each failure and the related changes I've made: http://people.apache.org/~niallp/io-2.0/IOFailures.html >From seeing Continuum failure emails, it may seem that it has been failing alot - but I think that the above shows that this is now not the case for the individual tests. Niall > Because of these changes I'm going to roll another RC > >> Niall >> >>> - Jörg >> > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
