paradedb / pg_analytics

DuckDB-powered analytics for Postgres
https://paradedb.com
PostgreSQL License
383 stars 15 forks source link

'time_bucket' function clashes with timescaledb #150

Closed ssidorenko closed 3 weeks ago

ssidorenko commented 1 month ago

What happens?

When creating extension on a server with timescaledb already created, it fails because the function name time_bucket is already taken.

To Reproduce

OS:

Linux x64 (Alma Linux 9.4)

ParadeDB Version:

0.1.4

Are you using ParadeDB Docker, Helm, or the extension(s) standalone?

ParadeDB pg_analytics Extension

Full Name:

Semion Sidorenko

Affiliation:

Sidorenko Consulting

Did you include all relevant data sets for reproducing the issue?

Yes

Did you include the code required to reproduce the issue?

Did you include all relevant configurations (e.g., CPU architecture, PostgreSQL version, Linux distribution) to reproduce the issue?

philippemnoel commented 1 month ago

Hey @ssidorenko! We haven't really invested effort in making pg_analytics compatible with Timescale. What is your use case?

If you'd like to submit a patch for this, we'd love. But otherwise I'm going to close this as a non-issue.

cobolbaby commented 1 month ago

Why install third-party extensions in the public schema? Is there a way to configure them to be installed in the default schema?

@philippemnoel

phyk commented 4 weeks ago

We have a use case where we want to have a delta table managed by pg_analytics on the one hand and streaming data ingested into the postgres on the other hand. From what I can see pg_analytics does not support a) streaming data efficiently and b) providing a way to do real time analytics as timescaledb does.

For that reason a combination of both would be beneficial?

phyk commented 4 weeks ago

Cross linked the issue here https://github.com/timescale/timescaledb/issues/7387

philippemnoel commented 4 weeks ago

Cross linked the issue here timescale/timescaledb#7387

Terrific, thank you!

philippemnoel commented 4 weeks ago

If this is required for Timescale compatibility, we're happy to rename the functions/remove them. I dont' think they're used actively today.

ssidorenko commented 3 weeks ago

We have a use case where we want to have a delta table managed by pg_analytics on the one hand and streaming data ingested into the postgres on the other hand. From what I can see pg_analytics does not support a) streaming data efficiently and b) providing a way to do real time analytics as timescaledb does.

For that reason a combination of both would be beneficial?

We had a very similar situation. For unrelated reasons, I ended up going for a different architecture that did not use pg_analytics, but at the time it sure would have been nice to be able to have both extensions in the same DB. I had to provision separate database instances instead one for timescale and one for pg_analytics, and use FDWs or Trino for cross-db querying.

philippemnoel commented 3 weeks ago

I removed the datetime function, since it wasn't being used.