Tx-report-queue

Client API
1

Is there intention to support the tx-report-queue in the Client-API?

We have built a live query streaming engine on top of Datomics Peer API that runs in all our customers projects and would like to port/test it out against the Client API. Also we usually have several dispatch loops running on the tx-report-queue for event triggering etc. If we were to port any running application or start a new one with the Client API, it seems like we’d need to have the tx-report-queue.

What are your thoughts/recommendations?

Kind regards, Leon.

1 Like
2

A simple indication whether this is going to happen or not would be very helpful. Please :slight_smile:

3

Apologies, but we do not publish a roadmap of Datomic feature development. As such, I cannot comment on the tx-report-queue feature request.

Pulling from another thread:

The product team has made a commitment to stewarding the product according to technical best practices and does not want to run the risk of customers developing against promised functionality that may turn out to be a bad idea for the product as a whole or reprioritized in favor of more critical features. We welcome and encourage technical feedback and feature requests. Those are taken seriously and made part of the product management decision-making.

I understand this might not be the most satisfying answer or the indication you were looking for, but I hope it gives you some insight as to why we do not pre-announce features.

Thanks,
Jaret

1 Like
4

Thats fine with me. Would be happy if you’d forward to the team my feedback that the Client API/Datomic Cloud is not useful in 5/6 projects without the tx-report queue.

5

I asked Rich about this at the party, he suggested that there may be other ways, for example if we are watching the tx stream to run effects, instead of that a transaction function could push to a queue

6

But tx fns are supposed to be pure right?

7

I think tx fns need to be idempotent but i don’t see why they need be pure.

8

Just wondering if anyone figured out an alternative approach for this problem? Or maybe someone from Cognitect can drop us a hint?

There is a tx-range API, maybe we can poll it in short interval?