Ignasi Marimon-Clos

Results 258 comments of Ignasi Marimon-Clos

I haven't jumped into this codebase in a while and realized that `logicalDelete` is only used in the (now legacy) `ByteArrayJournalDao`. The `ByteArrayJournalDao` class is not deprecated but was moved...

Sure, what I'm saying is that in a (hypothetical) version v6.0.0 of this plugin we would have the migration tool (from the legacy `ByteArrayJournalDao` to the `DefaultJournalDao`). Then, changing the...

Again, except this time the failure was on the H2 flavor of the test: (it was MySQL's on the case @chbatey reported above) https://travis-ci.com/github/akka/akka-persistence-jdbc/jobs/474105442#L960

Thanks @nvollmar for puting this in motion. This is a rather important behavioral change so, IMHO it should be prominently explained on the docs and release notes. Maybe even justifies...

In general, SQL queries must still filter by `logicalDeletion.enable = false` but maybe when the flag to hard-delete is enabled we can remove the filter by `journal.deleted=false` criteria from the...

Do we already have documentation on the safest procedure to move from `logicalDeletion.enable = false` to `logicalDeletion.enable = true` (and back in case of a rollback)? Maybe for a separate...

Thanks for checking. Maybe, since there's already a deprecation warning and changing the default value of the setting already requires a MAJOR version bump, maybe we could just drop the...

Of course, there are a million external conditionals that may be affecting this quick and dirty study (running the DB on docker on a local laptop, _small_ dataset of 150k...

*BONUS* the `order by x2."ordering"` clause uses the index in the journal table instead of using the index on the evetn_tag table. I've done some tests changing both the filter...

*MENTAL NOTE* I suspect the reason for the observed behavior is that the primary key on `event_tag` is defined as `PRIMARY KEY(event_id, tag)` (that is `event_id` first). I wonder if...