Sebastian Utz

👍 I really like this approach and would definitely prefer it over #437.

@proddata Thank you for this well described request. I've changed the title to reflect the outcome one want to archive instead of mentioning a possible implementation as you also mentioned...

> However, it seems automatic run doesn't use directly some internal code as I thought before, it calls ANALYZE statement > https://github.com/crate/crate/blob/master/server/src/main/java/io/crate/statistics/TableStatsService.java#L111 > >so we need extend grammar already -...

@BaurzhanSakhariev > I think we cannot refer to not yet merged doc? Of course you can, you can (and also please do) validate your documentation/CHANGES entry changes by building the...

@loredp Elasticsearch plugins are not officially supported but may work with small modifications. Our current `TokenizerFactory` API is based on Elasticsearch v6.8, such only analysis plugins for this version are...

@loredp We are also welcoming any contribution or maybe you want to create and host a CrateDB variant of that plugin on your own.

Doing a network lookup to the GeoIP2 (or GeoLite2) on every single written row could dramatically decrease performance. Also possible errors like timeout's, disconnects etc. must be handle safely. A...

> The idea is to not use the webservice. MaxMind has a downloadable database that you can perform lookups with. Multiple geolocation companies use this database format. > >https://github.com/maxmind/GeoIP2-java#database-usage Thanks...

An example for post-processing the exported JSONL to CSV using [jq](https://stedolan.github.io/jq/): ```jq -r -s '(map(keys) | add | unique) as $cols | map(. as $row | $cols | map($row[.])) as...

We should probably switch to extend from the `PGCompiler` instead of the default `SQLCompiler`. See https://github.com/sqlalchemy/sqlalchemy/blob/rel_1_4_41/lib/sqlalchemy/dialects/postgresql/base.py#L2394. CrateDB (like PG) does not support negative LIMIT values.