Re: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
Just got a successful build on Windows 10 JDK 11 after fixing line endings of testVCalendar.vcs +1 Em seg., 2 de mai. de 2022 às 11:33, Tim Allison escreveu: > > Maybe we should configure that explicitly in .gitattributes in the repo. > > +1 go for it! > > On Mon, May 2, 2022 at 10:11 AM Luís Filipe Nassif > wrote: > > > > You're right Tim! Those test files are using \r\n. I'll set EOL to \n, > > clone from scratch and try again. Maybe we should configure that > explicitly > > in .gitattributes in the repo. > > > > Em seg., 2 de mai. de 2022 às 10:44, Tim Allison > > escreveu: > > > > > Hahahaha...doh. I've gotten those kinds of errors before on Windows > > > when I hadn't set git global EOL to "\n". There's a chance that git > > > has converted the line endings in those test files to \r\n. Can you > > > open the test files in a hex editor and see what the line endings look > > > like? > > > > > > We need to improve documentation on line endings on Windows. > > > > > > On Mon, May 2, 2022 at 9:40 AM Luís Filipe Nassif > > > > wrote: > > > > > > > > Just got these build failures on Windows 10 JDK 11: > > > > > > > > [ERROR] Failures: > > > > [ERROR] TextAndCSVParserTest.testSubclassingMimeTypesRemain:217 > > > > expected:<...-vcalendar; charset=[ISO-8859-1]> but > was:<...-vcalendar; > > > > charset=[windows-1252]> > > > > [ERROR] TXTParserTest.testSubclassingMimeTypesRemain:299 > > > > expected:<...-vcalendar; charset=[ISO-8859-1]> but > was:<...-vcalendar; > > > > charset=[windows-1252]> > > > > > > > > > > > > Em seg., 2 de mai. de 2022 às 08:15, Tim Allison < > talli...@apache.org> > > > > escreveu: > > > > > > > > > Thank you, Tilman! > > > > > > > > > > I'll give it a few more hours in case anyone wants to -1 it. > > > > > > > > > > On Mon, May 2, 2022 at 6:20 AM Tilman Hausherr < > thaush...@t-online.de> > > > > > wrote: > > > > > > > > > > > > Go with what we have > > > > > > Tilman > > > > > > > > > > > > > > > > > > > > > > > > --- Original-Nachricht --- > > > > > > Von: Tim Allison > > > > > > Betreff: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2 > > > > > > Datum: 02. Mai 2022, 12:15 > > > > > > An: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > I confirmed my memory of TikaServerIntegrationTest. The > difference is > > > > > > that in 1.x, we were still using threads to start the server and > then > > > > > > hoping that<http://thread.interrupt> () actually shut the > process > > > down > > > > > in a > > > > > > reasonable amount of time. In 2.x, we're starting processes and > then > > > > > > force quitting those after each test. In my experience, this got > rid > > > > > > of the flaky tests in 2.x > > > > > > > > > > > > We didn't change any of the underlying server logic in 1.x. This > is a > > > > > > problem of flaky tests. > > > > > > > > > > > > Given that we have the 1.x branch through September, I'll > convert the > > > > > > thread based tests to process based tests shortly. The question > is: > > > > > > should we roll an rc3 or go with what we have. > > > > > > > > > > > > Thank you, all. > > > > > > > > > > > > Best, > > > > > > > > > > > > Tim > > > > > > > > > > > > On Sat, Apr 30, 2022 at 9:10 AM Tim Allison > > > > > <mailto:talli...@apache.org> > wrote: > > > > > > > > > > > > > > That's exactly my understanding as well. That test is not > flaky in > > > 2x > > > > > > > because of modifications I made to the integration tests in 2x. > > > > > > > > > > > > > > In 2x, the Solr tests can be flaky, and there's an open issue > for > > > > > > > that. I don't like flaky tests. Sorry. > >
Re: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
> Maybe we should configure that explicitly in .gitattributes in the repo. +1 go for it! On Mon, May 2, 2022 at 10:11 AM Luís Filipe Nassif wrote: > > You're right Tim! Those test files are using \r\n. I'll set EOL to \n, > clone from scratch and try again. Maybe we should configure that explicitly > in .gitattributes in the repo. > > Em seg., 2 de mai. de 2022 às 10:44, Tim Allison > escreveu: > > > Hahahaha...doh. I've gotten those kinds of errors before on Windows > > when I hadn't set git global EOL to "\n". There's a chance that git > > has converted the line endings in those test files to \r\n. Can you > > open the test files in a hex editor and see what the line endings look > > like? > > > > We need to improve documentation on line endings on Windows. > > > > On Mon, May 2, 2022 at 9:40 AM Luís Filipe Nassif > > wrote: > > > > > > Just got these build failures on Windows 10 JDK 11: > > > > > > [ERROR] Failures: > > > [ERROR] TextAndCSVParserTest.testSubclassingMimeTypesRemain:217 > > > expected:<...-vcalendar; charset=[ISO-8859-1]> but was:<...-vcalendar; > > > charset=[windows-1252]> > > > [ERROR] TXTParserTest.testSubclassingMimeTypesRemain:299 > > > expected:<...-vcalendar; charset=[ISO-8859-1]> but was:<...-vcalendar; > > > charset=[windows-1252]> > > > > > > > > > Em seg., 2 de mai. de 2022 às 08:15, Tim Allison > > > escreveu: > > > > > > > Thank you, Tilman! > > > > > > > > I'll give it a few more hours in case anyone wants to -1 it. > > > > > > > > On Mon, May 2, 2022 at 6:20 AM Tilman Hausherr > > > > wrote: > > > > > > > > > > Go with what we have > > > > > Tilman > > > > > > > > > > > > > > > > > > > > --- Original-Nachricht --- > > > > > Von: Tim Allison > > > > > Betreff: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2 > > > > > Datum: 02. Mai 2022, 12:15 > > > > > An: > > > > > > > > > > > > > > > > > > > > > > > > > I confirmed my memory of TikaServerIntegrationTest. The difference is > > > > > that in 1.x, we were still using threads to start the server and then > > > > > hoping that<http://thread.interrupt> () actually shut the process > > down > > > > in a > > > > > reasonable amount of time. In 2.x, we're starting processes and then > > > > > force quitting those after each test. In my experience, this got rid > > > > > of the flaky tests in 2.x > > > > > > > > > > We didn't change any of the underlying server logic in 1.x. This is a > > > > > problem of flaky tests. > > > > > > > > > > Given that we have the 1.x branch through September, I'll convert the > > > > > thread based tests to process based tests shortly. The question is: > > > > > should we roll an rc3 or go with what we have. > > > > > > > > > > Thank you, all. > > > > > > > > > > Best, > > > > > > > > > > Tim > > > > > > > > > > On Sat, Apr 30, 2022 at 9:10 AM Tim Allison > > > > <mailto:talli...@apache.org> > wrote: > > > > > > > > > > > > That's exactly my understanding as well. That test is not flaky in > > 2x > > > > > > because of modifications I made to the integration tests in 2x. > > > > > > > > > > > > In 2x, the Solr tests can be flaky, and there's an open issue for > > > > > > that. I don't like flaky tests. Sorry. > > > > > > > > > > > > On Sat, Apr 30, 2022 at 6:28 AM Tilman Hausherr < > > thaush...@t-online.de > > > > > <mailto:thaush...@t-online.de> > wrote: > > > > > > > > > > > > > > Now I had successful builds with jdk11 and jdk18. I think this is > > > > more > > > > > a > > > > > > > problem with the test than with the software. I remember when I > > > > started > > > > > > > with tika I wanted to make it possible to have tika1 and 2 work > > in > > > > > > > parallel and never managed to do it and then moved on to other > > > > things. > > > > > > > Something about unreliability of the server starting and > > stopping. > > > > > > > > > > > > > > Tilman > > > > > > > > > > > > > > Am 30.04.2022 um 11:29 schrieb Tilman Hausherr: > > > > > > > > [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time > > > > > > > > elapsed: 102.543 s <<< FAILURE! - in > > > > > > > ><http://org.apache.tika.server.TikaServerIntegrationTest> > > > > > > > > [ERROR] > > > > > > > > > > > > > < > > > > > > http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > > > > > > > > > > > > Time elapsed: 5.694 s <<< ERROR! > > > > > > > ><http://java.lang.IllegalStateException> : Not a JSON Object: > > null > > > > > > > > at > > > > > > > > > > > > > <http://org.apache.tika.server.TikaServerIntegrationTest.getServerId > > > > (TikaServerIntegrationTest.java:280> > > > > > ) > > > > > > > > > > > > > > > > at > > > > > > > > > > > > > < > > > > > > http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > > > (TikaServerIntegrationTest.java:208> > > > > > ) > > > > > > > > > > > > > > > > W10, jdk11, maven 3.8.5 > > > > > > > > > > > > > > > > > > > > > > > > > > > > >
Re: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
You're right Tim! Those test files are using \r\n. I'll set EOL to \n, clone from scratch and try again. Maybe we should configure that explicitly in .gitattributes in the repo. Em seg., 2 de mai. de 2022 às 10:44, Tim Allison escreveu: > Hahahaha...doh. I've gotten those kinds of errors before on Windows > when I hadn't set git global EOL to "\n". There's a chance that git > has converted the line endings in those test files to \r\n. Can you > open the test files in a hex editor and see what the line endings look > like? > > We need to improve documentation on line endings on Windows. > > On Mon, May 2, 2022 at 9:40 AM Luís Filipe Nassif > wrote: > > > > Just got these build failures on Windows 10 JDK 11: > > > > [ERROR] Failures: > > [ERROR] TextAndCSVParserTest.testSubclassingMimeTypesRemain:217 > > expected:<...-vcalendar; charset=[ISO-8859-1]> but was:<...-vcalendar; > > charset=[windows-1252]> > > [ERROR] TXTParserTest.testSubclassingMimeTypesRemain:299 > > expected:<...-vcalendar; charset=[ISO-8859-1]> but was:<...-vcalendar; > > charset=[windows-1252]> > > > > > > Em seg., 2 de mai. de 2022 às 08:15, Tim Allison > > escreveu: > > > > > Thank you, Tilman! > > > > > > I'll give it a few more hours in case anyone wants to -1 it. > > > > > > On Mon, May 2, 2022 at 6:20 AM Tilman Hausherr > > > wrote: > > > > > > > > Go with what we have > > > > Tilman > > > > > > > > > > > > > > > > --- Original-Nachricht --- > > > > Von: Tim Allison > > > > Betreff: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2 > > > > Datum: 02. Mai 2022, 12:15 > > > > An: > > > > > > > > > > > > > > > > > > > > I confirmed my memory of TikaServerIntegrationTest. The difference is > > > > that in 1.x, we were still using threads to start the server and then > > > > hoping that<http://thread.interrupt> () actually shut the process > down > > > in a > > > > reasonable amount of time. In 2.x, we're starting processes and then > > > > force quitting those after each test. In my experience, this got rid > > > > of the flaky tests in 2.x > > > > > > > > We didn't change any of the underlying server logic in 1.x. This is a > > > > problem of flaky tests. > > > > > > > > Given that we have the 1.x branch through September, I'll convert the > > > > thread based tests to process based tests shortly. The question is: > > > > should we roll an rc3 or go with what we have. > > > > > > > > Thank you, all. > > > > > > > > Best, > > > > > > > > Tim > > > > > > > > On Sat, Apr 30, 2022 at 9:10 AM Tim Allison > > > <mailto:talli...@apache.org> > wrote: > > > > > > > > > > That's exactly my understanding as well. That test is not flaky in > 2x > > > > > because of modifications I made to the integration tests in 2x. > > > > > > > > > > In 2x, the Solr tests can be flaky, and there's an open issue for > > > > > that. I don't like flaky tests. Sorry. > > > > > > > > > > On Sat, Apr 30, 2022 at 6:28 AM Tilman Hausherr < > thaush...@t-online.de > > > > <mailto:thaush...@t-online.de> > wrote: > > > > > > > > > > > > Now I had successful builds with jdk11 and jdk18. I think this is > > > more > > > > a > > > > > > problem with the test than with the software. I remember when I > > > started > > > > > > with tika I wanted to make it possible to have tika1 and 2 work > in > > > > > > parallel and never managed to do it and then moved on to other > > > things. > > > > > > Something about unreliability of the server starting and > stopping. > > > > > > > > > > > > Tilman > > > > > > > > > > > > Am 30.04.2022 um 11:29 schrieb Tilman Hausherr: > > > > > > > [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time > > > > > > > elapsed: 102.543 s <<< FAILURE! - in > > > > > > ><http://org.apache.tika.server.TikaServerIntegrationTest> > > > > > > > [ERROR] > > > > > > > > > > > < > > > > http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > > > > > > > > > > Time elapsed: 5.694 s <<< ERROR! > > > > > > ><http://java.lang.IllegalStateException> : Not a JSON Object: > null > > > > > > > at > > > > > > > > > > > <http://org.apache.tika.server.TikaServerIntegrationTest.getServerId > > > (TikaServerIntegrationTest.java:280> > > > > ) > > > > > > > > > > > > > > at > > > > > > > > > > > < > > > > http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > > (TikaServerIntegrationTest.java:208> > > > > ) > > > > > > > > > > > > > > W10, jdk11, maven 3.8.5 > > > > > > > > > > > > > > > > > > > > > > > >
Re: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
Hahahaha...doh. I've gotten those kinds of errors before on Windows when I hadn't set git global EOL to "\n". There's a chance that git has converted the line endings in those test files to \r\n. Can you open the test files in a hex editor and see what the line endings look like? We need to improve documentation on line endings on Windows. On Mon, May 2, 2022 at 9:40 AM Luís Filipe Nassif wrote: > > Just got these build failures on Windows 10 JDK 11: > > [ERROR] Failures: > [ERROR] TextAndCSVParserTest.testSubclassingMimeTypesRemain:217 > expected:<...-vcalendar; charset=[ISO-8859-1]> but was:<...-vcalendar; > charset=[windows-1252]> > [ERROR] TXTParserTest.testSubclassingMimeTypesRemain:299 > expected:<...-vcalendar; charset=[ISO-8859-1]> but was:<...-vcalendar; > charset=[windows-1252]> > > > Em seg., 2 de mai. de 2022 às 08:15, Tim Allison > escreveu: > > > Thank you, Tilman! > > > > I'll give it a few more hours in case anyone wants to -1 it. > > > > On Mon, May 2, 2022 at 6:20 AM Tilman Hausherr > > wrote: > > > > > > Go with what we have > > > Tilman > > > > > > > > > > > > --- Original-Nachricht --- > > > Von: Tim Allison > > > Betreff: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2 > > > Datum: 02. Mai 2022, 12:15 > > > An: > > > > > > > > > > > > > > > I confirmed my memory of TikaServerIntegrationTest. The difference is > > > that in 1.x, we were still using threads to start the server and then > > > hoping that<http://thread.interrupt> () actually shut the process down > > in a > > > reasonable amount of time. In 2.x, we're starting processes and then > > > force quitting those after each test. In my experience, this got rid > > > of the flaky tests in 2.x > > > > > > We didn't change any of the underlying server logic in 1.x. This is a > > > problem of flaky tests. > > > > > > Given that we have the 1.x branch through September, I'll convert the > > > thread based tests to process based tests shortly. The question is: > > > should we roll an rc3 or go with what we have. > > > > > > Thank you, all. > > > > > > Best, > > > > > > Tim > > > > > > On Sat, Apr 30, 2022 at 9:10 AM Tim Allison > > <mailto:talli...@apache.org> > wrote: > > > > > > > > That's exactly my understanding as well. That test is not flaky in 2x > > > > because of modifications I made to the integration tests in 2x. > > > > > > > > In 2x, the Solr tests can be flaky, and there's an open issue for > > > > that. I don't like flaky tests. Sorry. > > > > > > > > On Sat, Apr 30, 2022 at 6:28 AM Tilman Hausherr > > <mailto:thaush...@t-online.de> > wrote: > > > > > > > > > > Now I had successful builds with jdk11 and jdk18. I think this is > > more > > > a > > > > > problem with the test than with the software. I remember when I > > started > > > > > with tika I wanted to make it possible to have tika1 and 2 work in > > > > > parallel and never managed to do it and then moved on to other > > things. > > > > > Something about unreliability of the server starting and stopping. > > > > > > > > > > Tilman > > > > > > > > > > Am 30.04.2022 um 11:29 schrieb Tilman Hausherr: > > > > > > [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time > > > > > > elapsed: 102.543 s <<< FAILURE! - in > > > > > ><http://org.apache.tika.server.TikaServerIntegrationTest> > > > > > > [ERROR] > > > > > > > > > < > > http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > > > > > > > > Time elapsed: 5.694 s <<< ERROR! > > > > > ><http://java.lang.IllegalStateException> : Not a JSON Object: null > > > > > > at > > > > > > > > > <http://org.apache.tika.server.TikaServerIntegrationTest.getServerId > > (TikaServerIntegrationTest.java:280> > > > ) > > > > > > > > > > > > at > > > > > > > > > < > > http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > (TikaServerIntegrationTest.java:208> > > > ) > > > > > > > > > > > > W10, jdk11, maven 3.8.5 > > > > > > > > > > > > > > > > > > >
Re: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
Just got these build failures on Windows 10 JDK 11: [ERROR] Failures: [ERROR] TextAndCSVParserTest.testSubclassingMimeTypesRemain:217 expected:<...-vcalendar; charset=[ISO-8859-1]> but was:<...-vcalendar; charset=[windows-1252]> [ERROR] TXTParserTest.testSubclassingMimeTypesRemain:299 expected:<...-vcalendar; charset=[ISO-8859-1]> but was:<...-vcalendar; charset=[windows-1252]> Em seg., 2 de mai. de 2022 às 08:15, Tim Allison escreveu: > Thank you, Tilman! > > I'll give it a few more hours in case anyone wants to -1 it. > > On Mon, May 2, 2022 at 6:20 AM Tilman Hausherr > wrote: > > > > Go with what we have > > Tilman > > > > > > > > --- Original-Nachricht --- > > Von: Tim Allison > > Betreff: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2 > > Datum: 02. Mai 2022, 12:15 > > An: > > > > > > > > > > I confirmed my memory of TikaServerIntegrationTest. The difference is > > that in 1.x, we were still using threads to start the server and then > > hoping that<http://thread.interrupt> () actually shut the process down > in a > > reasonable amount of time. In 2.x, we're starting processes and then > > force quitting those after each test. In my experience, this got rid > > of the flaky tests in 2.x > > > > We didn't change any of the underlying server logic in 1.x. This is a > > problem of flaky tests. > > > > Given that we have the 1.x branch through September, I'll convert the > > thread based tests to process based tests shortly. The question is: > > should we roll an rc3 or go with what we have. > > > > Thank you, all. > > > > Best, > > > > Tim > > > > On Sat, Apr 30, 2022 at 9:10 AM Tim Allison > <mailto:talli...@apache.org> > wrote: > > > > > > That's exactly my understanding as well. That test is not flaky in 2x > > > because of modifications I made to the integration tests in 2x. > > > > > > In 2x, the Solr tests can be flaky, and there's an open issue for > > > that. I don't like flaky tests. Sorry. > > > > > > On Sat, Apr 30, 2022 at 6:28 AM Tilman Hausherr > <mailto:thaush...@t-online.de> > wrote: > > > > > > > > Now I had successful builds with jdk11 and jdk18. I think this is > more > > a > > > > problem with the test than with the software. I remember when I > started > > > > with tika I wanted to make it possible to have tika1 and 2 work in > > > > parallel and never managed to do it and then moved on to other > things. > > > > Something about unreliability of the server starting and stopping. > > > > > > > > Tilman > > > > > > > > Am 30.04.2022 um 11:29 schrieb Tilman Hausherr: > > > > > [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time > > > > > elapsed: 102.543 s <<< FAILURE! - in > > > > ><http://org.apache.tika.server.TikaServerIntegrationTest> > > > > > [ERROR] > > > > > > > < > http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > > > > > > Time elapsed: 5.694 s <<< ERROR! > > > > ><http://java.lang.IllegalStateException> : Not a JSON Object: null > > > > > at > > > > > > > <http://org.apache.tika.server.TikaServerIntegrationTest.getServerId > (TikaServerIntegrationTest.java:280> > > ) > > > > > > > > > > at > > > > > > > < > http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > (TikaServerIntegrationTest.java:208> > > ) > > > > > > > > > > W10, jdk11, maven 3.8.5 > > > > > > > > > > > > > > >
Re: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
Thank you, Tilman! I'll give it a few more hours in case anyone wants to -1 it. On Mon, May 2, 2022 at 6:20 AM Tilman Hausherr wrote: > > Go with what we have > Tilman > > > > --- Original-Nachricht --- > Von: Tim Allison > Betreff: Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2 > Datum: 02. Mai 2022, 12:15 > An: > > > > > I confirmed my memory of TikaServerIntegrationTest. The difference is > that in 1.x, we were still using threads to start the server and then > hoping that<http://thread.interrupt> () actually shut the process down in a > reasonable amount of time. In 2.x, we're starting processes and then > force quitting those after each test. In my experience, this got rid > of the flaky tests in 2.x > > We didn't change any of the underlying server logic in 1.x. This is a > problem of flaky tests. > > Given that we have the 1.x branch through September, I'll convert the > thread based tests to process based tests shortly. The question is: > should we roll an rc3 or go with what we have. > > Thank you, all. > > Best, > > Tim > > On Sat, Apr 30, 2022 at 9:10 AM Tim Allison <mailto:talli...@apache.org> > wrote: > > > > That's exactly my understanding as well. That test is not flaky in 2x > > because of modifications I made to the integration tests in 2x. > > > > In 2x, the Solr tests can be flaky, and there's an open issue for > > that. I don't like flaky tests. Sorry. > > > > On Sat, Apr 30, 2022 at 6:28 AM Tilman Hausherr <mailto:thaush...@t-online.de> > wrote: > > > > > > Now I had successful builds with jdk11 and jdk18. I think this is more > a > > > problem with the test than with the software. I remember when I started > > > with tika I wanted to make it possible to have tika1 and 2 work in > > > parallel and never managed to do it and then moved on to other things. > > > Something about unreliability of the server starting and stopping. > > > > > > Tilman > > > > > > Am 30.04.2022 um 11:29 schrieb Tilman Hausherr: > > > > [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time > > > > elapsed: 102.543 s <<< FAILURE! - in > > > ><http://org.apache.tika.server.TikaServerIntegrationTest> > > > > [ERROR] > > > > > <http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM> > > > > Time elapsed: 5.694 s <<< ERROR! > > > ><http://java.lang.IllegalStateException> : Not a JSON Object: null > > > > at > > > > > <http://org.apache.tika.server.TikaServerIntegrationTest.getServerId(TikaServerIntegrationTest.java:280> > ) > > > > > > > > at > > > > > <http://org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM(TikaServerIntegrationTest.java:208> > ) > > > > > > > > W10, jdk11, maven 3.8.5 > > > > > > > > > > >
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
I confirmed my memory of TikaServerIntegrationTest. The difference is that in 1.x, we were still using threads to start the server and then hoping that thread.interrupt() actually shut the process down in a reasonable amount of time. In 2.x, we're starting processes and then force quitting those after each test. In my experience, this got rid of the flaky tests in 2.x We didn't change any of the underlying server logic in 1.x. This is a problem of flaky tests. Given that we have the 1.x branch through September, I'll convert the thread based tests to process based tests shortly. The question is: should we roll an rc3 or go with what we have. Thank you, all. Best, Tim On Sat, Apr 30, 2022 at 9:10 AM Tim Allison wrote: > > That's exactly my understanding as well. That test is not flaky in 2x > because of modifications I made to the integration tests in 2x. > > In 2x, the Solr tests can be flaky, and there's an open issue for > that. I don't like flaky tests. Sorry. > > On Sat, Apr 30, 2022 at 6:28 AM Tilman Hausherr wrote: > > > > Now I had successful builds with jdk11 and jdk18. I think this is more a > > problem with the test than with the software. I remember when I started > > with tika I wanted to make it possible to have tika1 and 2 work in > > parallel and never managed to do it and then moved on to other things. > > Something about unreliability of the server starting and stopping. > > > > Tilman > > > > Am 30.04.2022 um 11:29 schrieb Tilman Hausherr: > > > [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time > > > elapsed: 102.543 s <<< FAILURE! - in > > > org.apache.tika.server.TikaServerIntegrationTest > > > [ERROR] > > > org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > > Time elapsed: 5.694 s <<< ERROR! > > > java.lang.IllegalStateException: Not a JSON Object: null > > > at > > > org.apache.tika.server.TikaServerIntegrationTest.getServerId(TikaServerIntegrationTest.java:280) > > > > > > at > > > org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM(TikaServerIntegrationTest.java:208) > > > > > > W10, jdk11, maven 3.8.5 > > > > > > > >
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
That's exactly my understanding as well. That test is not flaky in 2x because of modifications I made to the integration tests in 2x. In 2x, the Solr tests can be flaky, and there's an open issue for that. I don't like flaky tests. Sorry. On Sat, Apr 30, 2022 at 6:28 AM Tilman Hausherr wrote: > > Now I had successful builds with jdk11 and jdk18. I think this is more a > problem with the test than with the software. I remember when I started > with tika I wanted to make it possible to have tika1 and 2 work in > parallel and never managed to do it and then moved on to other things. > Something about unreliability of the server starting and stopping. > > Tilman > > Am 30.04.2022 um 11:29 schrieb Tilman Hausherr: > > [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time > > elapsed: 102.543 s <<< FAILURE! - in > > org.apache.tika.server.TikaServerIntegrationTest > > [ERROR] > > org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > > Time elapsed: 5.694 s <<< ERROR! > > java.lang.IllegalStateException: Not a JSON Object: null > > at > > org.apache.tika.server.TikaServerIntegrationTest.getServerId(TikaServerIntegrationTest.java:280) > > > > at > > org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM(TikaServerIntegrationTest.java:208) > > > > W10, jdk11, maven 3.8.5 > > > > >
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
Now I had successful builds with jdk11 and jdk18. I think this is more a problem with the test than with the software. I remember when I started with tika I wanted to make it possible to have tika1 and 2 work in parallel and never managed to do it and then moved on to other things. Something about unreliability of the server starting and stopping. Tilman Am 30.04.2022 um 11:29 schrieb Tilman Hausherr: [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time elapsed: 102.543 s <<< FAILURE! - in org.apache.tika.server.TikaServerIntegrationTest [ERROR] org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM Time elapsed: 5.694 s <<< ERROR! java.lang.IllegalStateException: Not a JSON Object: null at org.apache.tika.server.TikaServerIntegrationTest.getServerId(TikaServerIntegrationTest.java:280) at org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM(TikaServerIntegrationTest.java:208) W10, jdk11, maven 3.8.5
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
I have seen that test flake out. I’ll take a look on Monday. On Sat, Apr 30, 2022 at 5:30 AM Tilman Hausherr wrote: > [ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time elapsed: > 102.543 s <<< FAILURE! - in > org.apache.tika.server.TikaServerIntegrationTest > [ERROR] > org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM > Time elapsed: 5.694 s <<< ERROR! > java.lang.IllegalStateException: Not a JSON Object: null > at > > org.apache.tika.server.TikaServerIntegrationTest.getServerId(TikaServerIntegrationTest.java:280) > > at > > org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM(TikaServerIntegrationTest.java:208) > > W10, jdk11, maven 3.8.5 > > >
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
[ERROR] Tests run: 12, Failures: 0, Errors: 1, Skipped: 2, Time elapsed: 102.543 s <<< FAILURE! - in org.apache.tika.server.TikaServerIntegrationTest [ERROR] org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM Time elapsed: 5.694 s <<< ERROR! java.lang.IllegalStateException: Not a JSON Object: null at org.apache.tika.server.TikaServerIntegrationTest.getServerId(TikaServerIntegrationTest.java:280) at org.apache.tika.server.TikaServerIntegrationTest.testSameServerIdAfterOOM(TikaServerIntegrationTest.java:208) W10, jdk11, maven 3.8.5
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
Am 30.04.2022 um 11:08 schrieb Tilman Hausherr: Me, I did a successful build yesterday with jdk8 (oracle) and now with jdk11 (oracle) on Windows 10. However there are frequent build fails on the CI for jdk11 and jdk 17. https://ci-builds.apache.org/job/tika/job/tika-branch1x-jdk11/170/ Tilman Oops I just noticed that I had "-skipTests" in my jdk11 "script". I'll run it again and report. Tilman
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
Am 30.04.2022 um 10:32 schrieb David Meikle: Everything looks good (sigs, etc) and has build and tests fine for me in Ubuntu 22.04 (Java 11 and 17) but is failing on Windows 11 (Java 11). The test that is failing TikaServerIntegrationTest.testSameServerIdAfterOOM() is consistent in Maven build but not in IDE. Will drill into it today to see if I can see anything. Anyone building fine on Windows? Me, I did a successful build yesterday with jdk8 (oracle) and now with jdk11 (oracle) on Windows 10. However there are frequent build fails on the CI for jdk11 and jdk 17. https://ci-builds.apache.org/job/tika/job/tika-branch1x-jdk11/170/ Tilman
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
On Thu, 28 Apr 2022 at 15:55, Tim Allison wrote: > A candidate for the Tika 1.28.2 release is available at: > https://dist.apache.org/repos/dist/dev/tika/1.28.2 > > The release candidate is a zip archive of the sources in: > https://github.com/apache/tika/tree/1.28.2-rc2/ > > The SHA-512 checksum of the archive is > > 035f3643a302e2a88f99ca549c4d5c5c6eecd7736d03e4a686b17028f519f6a7a40229e48f2aac0bdf1653391e0bd7d34d0c7d099a2e5a2cb6141df00a4181bf. > > In addition, a staged maven repository is available here: > > https://repository.apache.org/content/repositories/orgapachetika-1083/org/apache/tika > > Please vote on releasing this package as Apache Tika 1.28.2. > The vote is open for the next 72 hours and passes if a majority of at > least three +1 Tika PMC votes are cast. > > [ ] +1 Release this package as Apache Tika 1.28.2 > [ ] -1 Do not release this package because... > > Everything looks good (sigs, etc) and has build and tests fine for me in Ubuntu 22.04 (Java 11 and 17) but is failing on Windows 11 (Java 11). The test that is failing TikaServerIntegrationTest.testSameServerIdAfterOOM() is consistent in Maven build but not in IDE. Will drill into it today to see if I can see anything. Anyone building fine on Windows? Cheers, Dave
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
Hi, +1. Basic stuff, linux ubuntu 20, x86, java 11. Thanks. On Thu, Apr 28, 2022, 20:23 Tilman Hausherr wrote: > +1 > > Tilman > > Am 28.04.2022 um 16:54 schrieb Tim Allison: > > A candidate for the Tika 1.28.2 release is available at: > >https://dist.apache.org/repos/dist/dev/tika/1.28.2 > > > > The release candidate is a zip archive of the sources in: > >https://github.com/apache/tika/tree/1.28.2-rc2/ > > > > The SHA-512 checksum of the archive is > > > 035f3643a302e2a88f99ca549c4d5c5c6eecd7736d03e4a686b17028f519f6a7a40229e48f2aac0bdf1653391e0bd7d34d0c7d099a2e5a2cb6141df00a4181bf. > > > > In addition, a staged maven repository is available here: > > > https://repository.apache.org/content/repositories/orgapachetika-1083/org/apache/tika > > > > Please vote on releasing this package as Apache Tika 1.28.2. > > The vote is open for the next 72 hours and passes if a majority of at > > least three +1 Tika PMC votes are cast. > > > > [ ] +1 Release this package as Apache Tika 1.28.2 > > [ ] -1 Do not release this package because... > > > > > > Here's my +1. > > > > Best, > > > > Tim > > >
Re: [VOTE] Release Apache Tika 1.28.2 Candidate #2
+1 Tilman Am 28.04.2022 um 16:54 schrieb Tim Allison: A candidate for the Tika 1.28.2 release is available at: https://dist.apache.org/repos/dist/dev/tika/1.28.2 The release candidate is a zip archive of the sources in: https://github.com/apache/tika/tree/1.28.2-rc2/ The SHA-512 checksum of the archive is 035f3643a302e2a88f99ca549c4d5c5c6eecd7736d03e4a686b17028f519f6a7a40229e48f2aac0bdf1653391e0bd7d34d0c7d099a2e5a2cb6141df00a4181bf. In addition, a staged maven repository is available here: https://repository.apache.org/content/repositories/orgapachetika-1083/org/apache/tika Please vote on releasing this package as Apache Tika 1.28.2. The vote is open for the next 72 hours and passes if a majority of at least three +1 Tika PMC votes are cast. [ ] +1 Release this package as Apache Tika 1.28.2 [ ] -1 Do not release this package because... Here's my +1. Best, Tim