Clojurians Log v2

For example, after this insert:

INSERT INTO locations (_id, owner)
VALUES (1, {id: 'foo'})

SELECT _id, owner.id
FROM locations

| _id |
|-----|
| 1   |

I can't remember the reason, but you have to put https://play.xtdb.com/?version=2.0.0-SNAPSHOT&type=sql&txs=W3sic3lzdGVtLXRpbWUiOm51bGwsInR4cyI6IklOU0VSVCBJTlRPIGxvY2F0aW9ucyAoX2lkLCBvd25lcilcbiBWQUxVRVMgKDEsIHtpZDogJ2Zvbyd9KSJ9XQ%3D%3D&query=U0VMRUNUIF9pZCwgKG93bmVyKS5pZCBGUk9NIGxvY2F0aW9ucw%3D%3D.

Thanks! Is that written anywhere in the docs?

I don't believe so, I seem to remember it came from the SQL standard. One to improve on either way!

Ah, found this issue that explains things: https://github.com/xtdb/xtdb/issues/3436

Hey @pafitisnick https://github.com/apache/arrow-adbc/issues/547 would be a good upstream ADBC drive issue to track / subscribe to Longer term though I don't know if r2dbc has a meaningful role if everyone switches to using virtual threads - looking at https://mariadb.com/resources/blog/benchmark-jdbc-connectors-and-java-21-virtual-threads/ > [...] in terms of performance, virtual threads dethrone R2DBC

How would vthreads help with reactivity though?

ah right, I was thinking about async more generally - but I guess you mean more like for listening to tx-log changes? or if we were to implement Incremental View Maintenance?

I'm thinking of queries that efficiently listen to changes

a reactive tx-log listener query like SELECT * FROM _txs ORDER BY _id ASC would be viable in the not-so-distant future, but what other kinds of queries are you hoping to be efficiently reactive? Anything that involves a join is like not possible without introducing an IVM engine (or if you can tolerate some latency then batch materialized view recalculations could be useful here also)

I think I misunderstood r2dbc

I'm just trying to find a database that I can use in my hyperfiddle/electric apps, that would allow to make a query and stream any changes, thus be able to make highly interactive realtime webapps

It's a trend, where a lot of databases support "reactive" (or "live", "real-time") queries, an example is retthink, pocketbase, convex, surreal etc. Ideally for me I'd want to use XTDB, for all the other features XTDB provides, (bitemporality, graph queries, edn native and so on)

This is another one of those nested columns not being accessed correctly issues.

Looks like you're running afoul of that nested column access syntax again. Putting brackets around (pl.partner) https://play.xtdb.com/?version=2.0.0-SNAPSHOT&type=sql&txs=W3sic3lzdGVtLXRpbWUiOm51bGwsInR4cyI6IklOU0VSVCBJTlRPIHBhcnRuZXJzIChfaWQsIG5hbWUpIFZBTFVFUyAoMSwgJ0FwdGVyYScpIn0seyJ0eHMiOiJJTlNFUlQgSU5UTyBwYXJ0bmVyX2xvY2F0aW9ucyAoX2lkLCBwYXJ0bmVyLCBuYW1lKVxuVkFMVUVTICgxLCB7aWQ6IDF9LCAnQ2FybHNiYWQnKSIsInN5c3RlbS10aW1lIjpudWxsfV0%3D&query=c2VsZWN0IHAubmFtZSwgcGwubmFtZSBhcyBsb2NhdGlvbl9uYW1lCmZyb20gcGFydG5lcnMgcAogIGxlZnQgb3V0ZXIgam9pbiBwYXJ0bmVyX2xvY2F0aW9ucyBwbCBvbiAocGwucGFydG5lcikuaWQgPSBwLl9pZA%3D%3D

Jinx

Thank you!

I’m running first two insert statements:

INSERT INTO partners (_id, name)
VALUES (1, 'Aptera')

INSERT INTO partner_locations (_id, partner, name)
VALUES (1, {id: 1}, 'Carlsbad')

select p.name, pl.name as location_name
from partners p
  left outer join partner_locations pl on pl.partner.id = p._id

https://play.xtdb.com/?version=2.0.0-SNAPSHOT&type=sql&txs=W3sic3lzdGVtLXRpbWUiOm51bGwsInR4cyI6IklOU0VSVCBJTlRPIHBhcnRuZXJzIChfaWQsIG5hbWUpIFZBTFVFUyAoMSwgJ0FwdGVyYScpIn0seyJ0eHMiOiJJTlNFUlQgSU5UTyBwYXJ0bmVyX2xvY2F0aW9ucyAoX2lkLCBwYXJ0bmVyLCBuYW1lKVxuVkFMVUVTICgxLCB7aWQ6IDF9LCAnQ2FybHNiYWQnKSIsInN5c3RlbS10aW1lIjpudWxsfV0%3D&query=c2VsZWN0IHAubmFtZSwgcGwubmFtZSBhcyBsb2NhdGlvbl9uYW1lCmZyb20gcGFydG5lcnMgcAogIGxlZnQgb3V0ZXIgam9pbiBwYXJ0bmVyX2xvY2F0aW9ucyBwbCBvbiBwbC5wYXJ0bmVyLmlkID0gcC5faWQ%3D