Re: Failures under JDK 9, 10, 11
I think even if we next support 11 as a minimum version, it's still useful to test with the intermediate versions since our users are using them now. On Tue, Nov 6, 2018 at 11:34 AM Erick Erickson wrote: > I didn't mean to suggest we switch to 11 as a minimum now, rather I > meant to say "If the next supported version will be 11 (whenever it's > stable enough), does it make sense to stop testing 9 and 10 now?". > > A better question would have been "do we ever expect to have Java 9 or > 10 as the minimum stable version? If not should we stop testing with > them?" > > Up to you of course. > On Tue, Nov 6, 2018 at 8:19 AM Uwe Schindler wrote: > > > > Hi, > > > Do we have any plans to release any Solr with minimal versions 9 or > > > 10? I'm wondering if it makes sense to stop testing 9 and 10 and plan > > > on the next supported Java version being 11 (whenever we do that). > > > > I don't think, we should now switch to Java 11 as minimum version yet. > I'd propose to do this after release of Lucene 8 (once branch_8x is > created) and only do that in the master branch. Of course, we can leave out > Java 9 and 10 and jump to 11. > > > > But interestingly: Java 9 and Java 10 are as stable as Java 8 (approx > same number of failures). Java 11 caused many more failures in Solr because > of some changes in TLS infrastructure (Java's support for TLS 1.3). We may > need to work on those problems. > > > > > > reduce the noise for failed tests from 9 and 10 > > > > repurpose those runs for more testing of 8 or 11 or 12 > > > > > > I don't have any strong feelings either way, it just popped into my > > > head and I thought I'd ask. > > > > See above, 9 and 10 are as satbel as 8, it's Java 11 and 12 that cause > more noise (of course this does not count crashes in some JVM versions). > > > > Uwe > > > > > > - > > To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org > > For additional commands, e-mail: dev-h...@lucene.apache.org > > > > - > To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org > For additional commands, e-mail: dev-h...@lucene.apache.org > > -- Lucene/Solr Search Committer, Consultant, Developer, Author, Speaker LinkedIn: http://linkedin.com/in/davidwsmiley | Book: http://www.solrenterprisesearchserver.com
Re: Failures under JDK 9, 10, 11
I didn't mean to suggest we switch to 11 as a minimum now, rather I meant to say "If the next supported version will be 11 (whenever it's stable enough), does it make sense to stop testing 9 and 10 now?". A better question would have been "do we ever expect to have Java 9 or 10 as the minimum stable version? If not should we stop testing with them?" Up to you of course. On Tue, Nov 6, 2018 at 8:19 AM Uwe Schindler wrote: > > Hi, > > Do we have any plans to release any Solr with minimal versions 9 or > > 10? I'm wondering if it makes sense to stop testing 9 and 10 and plan > > on the next supported Java version being 11 (whenever we do that). > > I don't think, we should now switch to Java 11 as minimum version yet. I'd > propose to do this after release of Lucene 8 (once branch_8x is created) and > only do that in the master branch. Of course, we can leave out Java 9 and 10 > and jump to 11. > > But interestingly: Java 9 and Java 10 are as stable as Java 8 (approx same > number of failures). Java 11 caused many more failures in Solr because of > some changes in TLS infrastructure (Java's support for TLS 1.3). We may need > to work on those problems. > > > > reduce the noise for failed tests from 9 and 10 > > > repurpose those runs for more testing of 8 or 11 or 12 > > > > I don't have any strong feelings either way, it just popped into my > > head and I thought I'd ask. > > See above, 9 and 10 are as satbel as 8, it's Java 11 and 12 that cause more > noise (of course this does not count crashes in some JVM versions). > > Uwe > > > - > To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org > For additional commands, e-mail: dev-h...@lucene.apache.org > - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
RE: Failures under JDK 9, 10, 11
Hi, > Do we have any plans to release any Solr with minimal versions 9 or > 10? I'm wondering if it makes sense to stop testing 9 and 10 and plan > on the next supported Java version being 11 (whenever we do that). I don't think, we should now switch to Java 11 as minimum version yet. I'd propose to do this after release of Lucene 8 (once branch_8x is created) and only do that in the master branch. Of course, we can leave out Java 9 and 10 and jump to 11. But interestingly: Java 9 and Java 10 are as stable as Java 8 (approx same number of failures). Java 11 caused many more failures in Solr because of some changes in TLS infrastructure (Java's support for TLS 1.3). We may need to work on those problems. > > reduce the noise for failed tests from 9 and 10 > > repurpose those runs for more testing of 8 or 11 or 12 > > I don't have any strong feelings either way, it just popped into my > head and I thought I'd ask. See above, 9 and 10 are as satbel as 8, it's Java 11 and 12 that cause more noise (of course this does not count crashes in some JVM versions). Uwe - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
