Roger Yang
Roger Yang
> I would agree with this if in-memory was actually used by end-users. It's a permutation that we support. Otherwise we should include postgres if faithfulness is paramount. But ultimately...
Also, using a disk-based database here is testing a few extra things: whether the disk is accessible, whether in-memory sqlite is different from disk-based sqlite. I think these extra "tests"...
related https://github.com/Arize-ai/openinference/issues/2284 These are non-message objects, and the OpenInference semantic convention currently can accommodate only message objects.
fyi. openai response payload ```json { "id": "resp_004a7bb90d2c503b0069161e5eeee0819d9b30518b082b0d91", "created_at": 1763057246, "error": null, "incomplete_details": null, "instructions": null, "metadata": {}, "model": "gpt-4.1-2025-04-14", "object": "response", "output": [ { "id": "ws_004a7bb90d2c503b0069161e5fbd10819d88910842a2e84a50", "action": { "query":...
fyi. corresponding langchain output ```json { "generations": [ [ { "text": "Madrid offers a vibrant array of events throughout November 2025, catering to diverse interests. Here are some highlights:\n\n\n\n**JazzMadrid International...
Can you please add a test similar to this [one](https://github.com/Arize-ai/openinference/blob/c3ce79402a05e99640b905274a16a8c0af8c04a2/python/instrumentation/openinference-instrumentation-langchain/tests/test_instrumentor.py#L65), and make sure it's working with concurrency? Thanks!
[fwiw](https://github.com/Arize-ai/phoenix/blob/849fa8b13ed003f3c4b1bdb326ac739d7aa8c87e/src/phoenix/trace/dsl/query.py#L771)
OK I now understand the issue. Thank you for bringing this to our attention. Let me discuss it with the team first, and follow up with you afterwards.
Thanks for the PR @Nynxz. Let me look into this issue a bit and get back to you.