[hibernate-dev] NoORM IRC meeting minutes

Tue May 21 09:42:29 EDT 2019

Hi,

Here are the minutes of this week's meeting (no bot for me so copy/paste
FTW):

15:04 < gsmet> #startmeeting
15:04 < gsmet> not having the bot there really starts to be an issue
15:04 < fax4ever> :/
15:04 < gsmet> anyway, let's start
15:04 < gsmet> #topic Progress Davide
15:04 < ddalto> The most interesting thing that happened is that I went to
London and talk with Mark Paluch (Spring Data Project Lead
15:04 < ddalto> ) and showed him the Hibernate Rx API.
15:05 < ddalto> Mark pointed out a couple of things that are a bit
inconsistent and might be worth to have a look.
15:05 < ddalto> I also saw his presentation on r2dbc (Reactive Relational
Database Connectivity) where he focused on the differences between
async/reactive
15:06 < ddalto> Other than that I continue to work on the reactive API
15:06 < fax4ever> nice!
15:07 -!- chammer [~
chatzilla at p200300C4BF0C4E00E9728938CEEE8926.dip0.t-ipconnect.de] has joined
#hibernate-dev
15:07 < chammer> eclipse-dev
15:07 < ddalto> That's mostly my sprint
15:07 < gsmet> ok, thanks
15:08 < gsmet> #topic Next 2 weeks Davide
15:09 < ddalto> I'm going to spend some time studying the r2dbc project
15:09 < ddalto> and see if I can adapt the API to be more consistent
15:09 < ddalto> I also have a PR on OGM that I need to look
15:10 < ddalto> That's all
15:10 < gsmet> ok, thanks
15:10 < gsmet> #topic Progress Fabio
15:10 < fax4ever> We skiped the meeting last time.
15:10 < fax4ever> So the progess this time will be about the last 4 weeks
of activity.
15:10 < fax4ever> Firstly, we fixed some pending pull requests,
15:10 < fax4ever> such as the one to check the analyzer compatibility
15:11 < fax4ever> for multi-index-full-text fields.
15:11 < fax4ever> Secondly, we did the issue about ElasticSearch schema
validation.
15:11 < fax4ever> In our tests we cover now all supported attributes.
15:11 < fax4ever> In the process we removed some attributes, like boost and
fieldData,
15:11 < fax4ever> and we discovered that the default of 'doc_values' from
15:11 < fax4ever> Elasticsearch 5 is really 'true' as declared in
Elasticsearch doc.
15:11 < fax4ever> Thirdly, we applied the new the built-in normalize method
15:11 < fax4ever> provided by Lucene whenever we need a normalization.
15:12 < fax4ever> The trick here was to override some methods in
TokenizerChain class.
15:12 < fax4ever> It was necessary because we don't use the Lucene-provided
CustomAnalyzer builder.
15:12 < fax4ever> Fourtly, we made thread safe
LuceneDistanceToFieldProjection class.
15:12 < fax4ever> The query builder cannot change the instance anymore,
15:12 < fax4ever> it only can change its own context.
15:12 < fax4ever> Fifthly, we avoided for any case the use of the Hibernate
Search session,
15:12 < fax4ever> if the underlying Hibernate ORM session is closed.
15:13 < fax4ever> if the underlying Hibernate ORM session is closed.
15:13 < fax4ever> Sixthly, I reviewed several pull requests, most of them
were quite extensive.
15:13 < fax4ever> Finally, the last two weeks were most about supporting
15:13 < fax4ever> the new annotation @DecimalScale and BigDecimal and
BigInteger backend field type.
15:13 < fax4ever> The rest of the time I reviewed further PRs.
15:13 < fax4ever> That's alla about the progress thanks
15:14 < gsmet> #topic Next 2 weeks Fabio
15:14 < fax4ever> I'm finishing the big pull request to support BigInteger
BigDecimal and scaling factor in general
15:14 < fax4ever> I'm almost to finish
15:15 < fax4ever> after I'll follow the lineup for the new sprint
15:15 < fax4ever> as usual
15:15 < fax4ever> so you can see our progress on the sprint page of jira
15:15 < fax4ever> that's all from me thanks
15:15 < gsmet> thanks
15:15 < gsmet> #topic Progress Guillaume
15:16 < gsmet> not much progress from my side
15:16 < gsmet> I reviewed and included a couple of Validator PRs
15:16 < gsmet> we had quite a few reports lately
15:16 < gsmet> Marko is fixing most of the bugs
15:16 < gsmet> we also had a significant external contributions which is
nice
15:17 < gsmet> other than that, I'm bugging Yoann about various things,
while I'm working on a Hibernate Search quickstart for Quarkus
15:17 < gsmet> hopefully, it's for the best but I admit I'm whining a lot :)
15:17 < fax4ever> cool!
15:17 < yrodiere> :)
15:17 < gsmet> #topic Next 2 weeks Guillaume
15:18 < gsmet> so I want to finish the guide + quickstart for Quarkus' next
release
15:18 < gsmet> I also have to update to the latest snapshots - next alpha
15:19 < gsmet> I also would like to release a new 6.0.x and a new
6.1.0.alpha for HV
15:19 < gsmet> that's it for me
15:19 < gsmet> #topic Progress Koen
15:19 < koentsje> i have been improving the create project wizard for the
quarkus tools
15:20 < koentsje> also continue the work on the cheat sheet
15:20 < koentsje> and i have tried to get continuous integration working
without much success, so i will definitely have to talk to yrodiere for this
15:21 < yrodiere> sure
15:21 < koentsje> but i did manage to set up a reddeer integration test
plugin
15:21 < koentsje> additionally i have released hibernate tools 5.3.10
15:22 < koentsje> and i have integrated the new release in the eclipse
tooling
15:22 < koentsje> and finally i have been working on integrating quarkus
0.15.0 into the eclipse tools
15:22 < koentsje> this is almost done, but still one or two loose ends
15:23 -!- sfikes [~sfikes at c-69-246-28-50.hsd1.ms.comcast.net] has joined
#hibernate-dev
15:23 < koentsje> that’s it
15:23 < gsmet> #topic Next 2 weeks Koen
15:23 < koentsje> first thing is finish the 0.15.0 integration
15:24 < koentsje> i also want to create a quarkus configuration for
running/debugging quarkus applications in eclipse
15:24 < koentsje> and improve the quarkus extensions view
15:25 < koentsje> and then, as said, investigate/implement continuous
integration
15:25 < koentsje> plus elaborate the reddeer automatic integration tests
15:25 < koentsje> if any time left, i want to investigate gradle for the
hibernate tools build
15:25 < koentsje> that will be it
15:26 < gsmet> ok, thanks
15:26 < gsmet> #topic Progress Yoann
15:26 < yrodiere> It's been a month for me too, so bear with me...
15:26 < yrodiere> First I sent a PR to add new APIs to declare dependencies
in bridges
15:26 < yrodiere> It was quite complicated to write, and probably even
worse to review (sorry Fabio!), but there are tests and in the end it got
merged
15:26 < yrodiere> I also upgraded Search 6 to ES 6.7.1 and 7
15:27 < yrodiere> Not a lot of changes there, just a bit of work to get rid
of some warnings
15:27 < yrodiere> That, and Fabio's work, got into the Search 6.0.0.Alpha5
release
15:27 < yrodiere> which was supposed to go into Quarkus 0.15, and maybe it
did, but we'll do another release before the Hibernate Search extension is
really advertised in Quarkus 0.16
15:27 < yrodiere> Then I worked on allowing users to configure the
"synchronization strategy" for automatic indexing in the Hibernate Search
ORM mapper
15:27 -!- sfikes [~sfikes at c-69-246-28-50.hsd1.ms.comcast.net] has quit
[Client Quit]
15:27 < yrodiere> This strategy basically describes how much to wait for
the indexing upon transaction commits
15:28 < yrodiere> There are three built-in strategies: "queued" (~ do not
wait), "committed" (~ wait until changes are persisted to disk, but don't
wait for them to be searchable) and "searchable" (you get the
                  idea)
15:28 < yrodiere> And there's also a way to plug in a custom strategy.
15:28 < yrodiere> The past two weeks, I fixed a bunch of tickets that were
really just waiting to be closed and were already fixed.
15:28 < yrodiere> I also submitted a PR to restore orchestration in the
Lucene backend
15:28 < yrodiere> Previously it was trivial, just serializing everything,
but that was supposed to be temporary
15:28 < yrodiere> As a result of this works the Lucene ITs went down from 2
minutes to 30s, which is nice
15:29 < fax4ever> ;)
15:29 < yrodiere> I suspect there is room for improvement, but at least
it's better than it used to be
15:29 < yrodiere> Beside that, I did some work on the Search DSL (again),
to make sure backends can fully extend it, down to the type of query and of
SearchResult.
15:29 < yrodiere> That's mainly to lay the groundwork for the introduction
of major backend-specific features, such as aggregations, which will
require a whole part of the DSL and probably a specific result
                  type too.
15:29 -!- sfikes [~sfikes at c-69-246-28-50.hsd1.ms.comcast.net] has joined
#hibernate-dev
15:29 < yrodiere> For good measure I introduced APIs for explaining query
score computation, which are different in Lucene and Elasticsearch: it
worked fine.
15:30 < yrodiere> I also changed the index schema declaration API a little
bit to force users to declare multi-valued fields, because we'll probably
need that information in the future.
15:30 < yrodiere> Finally..
15:30 < yrodiere> Yesterday and today, I worked on making the Search DSL
even less verbose by making some method calls optional
15:30 < yrodiere> It looks like this now:
15:30 < yrodiere> return Search.getSearchSession(em).search(Book.class)
15:30 < yrodiere>         .predicate(f ->
f.simpleQueryString().onFields("title").matching(pattern))
15:30 < yrodiere>         .sort(f -> f.byField("title_sort"))
15:30 < yrodiere>         .fetchHits();
15:30 < yrodiere> Well it's not merged yet, but close.
15:30 < yrodiere> Next two weeks!
15:30 < gsmet> #topic Next 2 weeks Yoann
15:31 < yrodiere> There's a big PRs still pending, so I expect to have to
do some work to do based on the reviews
15:31 < yrodiere> I will also have to release Search 6.0.0.Alpha6 for
Quarkus 0.16
15:31 < yrodiere> And ideally I'd like to release some of the maintenance
branches, too... but I've been postponing these for so long... who knows?
15:31 < yrodiere> After that, I will try to make Hibernate Search work in
the modulepath.
15:31 < yrodiere> I started to look into it, and it looks possible, but
there are a few problems with service loading that I suspect are related to
the aggregatedClassLoader in ORM. I still have to investigate.
15:32 < yrodiere> The problem is mainly that we already advertised this as
something that "just works", and I discovered some time ago that no, it
definitely did not
15:32 < yrodiere> so I'd like to fix that
15:32 < yrodiere> There are a few other tickets, but the main one will be
about restoring APIs for per-class works (purge, flush, ...) in the ORM
mapper. For now they are not exposed to end-users in Search 6,
                  they are just SPIs.
15:32 < yrodiere> That's all from me!
15:33 < gsmet> #topic Additional topics
15:33 < gsmet> I will be attending Mix-IT on Thursday-Friday so probably
not very active
15:34 < gsmet> RH is the main sponsor of the conference so we need people
there
15:34 < yrodiere> When is the Quarkus release again?
15:34 < gsmet> I won't be able to organize the hangout so Yoann, you're
hereby nominated chair :)
15:34 < gsmet> Wed 29th
15:34 < yrodiere> sure
15:35 < yrodiere> ok, so you'll be able to try out the Search release on
monday/tuesday
15:35 < fax4ever> :)
15:35 < yrodiere> well, have fun :)
15:35 < gsmet> well, we won't have people on the stand all day as there are
a lot of talks
15:35 < gsmet> so I'll be working as time permits
15:36 < gsmet> I really want to test this lazy thing before you release
15:36 < gsmet> anyone has anything to add?
15:36 < yrodiere> nope
15:36 < fax4ever> not from me thanks
15:36 < gsmet> OK, so let's close the meeting and have some fun
copy/pasting to an email...
15:37 < gsmet> #endmeeting

Cheers,

-- 
Guillaume