Going out on a limb here, as in “this may be a whacky idea but let’s toss it around anyway”.
- Check in the test fixes and Gradle build in the current state to master/9.0 as soon as possible. Essentially let it break however it will, even if that means we shut off Jenkins for a while. Or only test Lucene. Or….. - Concentrate (all of us in our copious spare time) on fixing master at least back to the state it is today reliability-wise. - Cut Solr 9.0 when it’s ready. My _guess_ would be early next year. - Slow/stop 8x development as necessary. - Yes, the 8x code line will have a short life, about 1 year. Given that 9.0 was the first one we required Java 11, I pretty much expected that. I don’t have strong feelings about this proposal, so feel free to shoot it down. That said, if we get the work Mark has done in front of everyone, there’s less incentive to “just let Mark do it”. And it’s not like we’d be breaking the currently released version. Erick > On Oct 21, 2019, at 7:39 AM, Martin Gainty <mgai...@hotmail.com> wrote: > > agree with david > as your schedule gets packed there are more than a few that are willing to > pickup the slack > if the branch is checked-in and labelled we can tackle what needs to be done > From: David Smiley <david.w.smi...@gmail.com> > Sent: Monday, October 21, 2019 2:13 AM > To: Mark Miller <markrmil...@gmail.com> > Cc: Lucene Dev <dev@lucene.apache.org> > Subject: Re: The Lucene Solr Gradle Build Game plan > > Thanks for your efforts Mark! > > Why has the migration to Gradle somehow been blocked by performance > improvements to Solr's tests? > What are the blockers to the Gradle branch being merged to master? It's an > experimental shadow build, so... none? > > I appreciate you want everything to be great from the beginning but I think > it's better to try and find that subjective line where it's good enough to be > committed then iterated on in follow-up issues. > > ~ David Smiley > Apache Lucene/Solr Search Developer > http://www.linkedin.com/in/davidwsmiley > > > On Mon, Oct 21, 2019 at 1:30 AM Mark Miller <markrmil...@gmail.com> wrote: > Hello Dawid, I'm sorry about the absence. I had a two week trip to SF and > that tends to steal my mind, thank god something does. > > Yes, I have local changes that you should likely wait for. > > There is a substantial issue with how I was mapping our directory names to > project names. That issue led to allowing some funny things and so addressing > it makes the build a bit to a lot less silly. Sorry Tomoko Uchida :( > > Anyway, I had to pause Gradle when I remembered that Solr tests are only slow > for silly or broken reasons. > > So I worked on that and then I remember that about a year and a half ago I > had realized SolrCloud was seriously rotting in come very core and key areas. > At the time I spent a lot of effort on my Starburst branch addressing some > fundamental problems and patterns. I then had a 6 week sabbatical and 3 week > movie making trip. Though my own lack of care, I lost all that work and > mostly forgot about the whole thing. > > Coming back to that, after wasting a lot of effort trying to find my previous > work, I set off to duplicate that work and hopefully take it a bit further. I > really can't work on anything unless until that is on a clear path to being > addressed unless someone attached to my livelihood forces me. > > Meanwhile, I'll be gone half of November on across the world vacation and > December is always a mess. > > On the plus side, the gradle build has gotten some great testing, but I'm > little fuzzy on when I'm ready. I will certainly be separating out these > efforts. So I'll be extracting the gradle changes to the gradle branch and > keeping my new state as a separate branch. > > - Mark --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
