Thank you, Dominik! For the regression tests and for fixing #1 already. As for #2, I wonder if we should look into using the actual Apache xerces for the next beta? Not another dependency....
-----Original Message----- From: Dominik Stadler [mailto:dominik.stad...@gmx.at] Sent: Friday, April 01, 2016 4:13 PM To: POI Developers List <dev@poi.apache.org> Subject: Re: Beta time yet? Hi, the POI specific regression test report is now updated for comparing 3.14-beta2 (nearly 3.14 final) with the current 3.15-beta1 (as of Tuesday) at http://people.apache.org/~centic/poi_regression/reports/ Runtime-wise I see that the run did take aprox 25h:15min now compared to 25h before, so I would say no actual change in runtime as I was doing other stuff on the machine for some time and thus might have affected these timings somewhat! I see two issues: 1. A XSLF related new error: java.lang.ArrayStoreException at java.util.ArrayList.toArray(ArrayList.java:408) at org.openxmlformats.schemas.drawingml.x2006.main.impl.CTGeomGuideListImpl.getGdArray(Unknown Source) at o.a.p.xslf.usermodel.XSLFSimpleShape.getAdjustValue(XSLFSimpleShape.java:823) Seems like list members with different types are used here so that putting them all into an array fails. However I could not reproduce this locally, so not sure how it was triggered in the regression testing?!? 2. An XML parsing related one: Caused by: java.lang.ArrayIndexOutOfBoundsException: 8192 at com.sun.org.apache.xerces.internal.impl.io.UTF8Reader.read(UTF8Reader.java:329) This seems to be a bug in the JDK itself in relation to surrogate Unicode characters, see e.g. https://bugs.openjdk.java.net/browse/JDK-7156085 and originally https://issues.apache.org/jira/browse/XERCESJ-1257 for more detailed discussion. Seems only JDK 9 has a fix for this :( However it is very rare, only 6 times in 1 mio documents, so I think it outweights the gain from using the JDK XML Parser. Dominik. On Wed, Mar 30, 2016 at 4:52 PM, Dominik Stadler <dominik.stad...@gmx.at> wrote: > > Ok, I'll try to squeeze in a run of my version of the large regression > test-suite with the latest trunk as it may cover a few different areas > not used in Tika. > > Dominik. > > On Wed, Mar 30, 2016 at 1:36 PM, kiwiwings <kiwiwi...@apache.org> wrote: > >> Nick Burch-8 wrote >> > What do people think about doing a beta release in a few weeks time ...? >> >> Last time it took ages between the thread start and the actual release. >> As the general consensus is anyway having short release cycles, I'll >> simply start the releasing trip on the next weekend (Friday evening). >> >> Maybe I should spent some time to script more of the manual work ... >> - as we use Jenkins here at my $dayjob, I'm used to (nearly) >> one-click-releasing the artifacts - I guess something like this would >> be possible with Ant too ... >> >> Andi >> >> >> >> -- >> View this message in context: >> http://apache-poi.1045710.n5.nabble.com/Beta-time-yet-tp5722477p57224 >> 81.html Sent from the POI - Dev mailing list archive at Nabble.com. >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@poi.apache.org For additional >> commands, e-mail: dev-h...@poi.apache.org >> >> > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@poi.apache.org For additional commands, e-mail: dev-h...@poi.apache.org
