Docs: Add disclaimer for static-parser when using execution mode local

[komjera]

Description

I was trying to optimize overhead for small models and saw that "--no-static-parser" was appended to all runs.

After some digging I found a comment in the source code stating that static-parser is not available in this execution mode. This should be disclosed in the documentation. I suggest adding a disclaimer here.

Use case/motivation

Transparency of this limitation would have saved me time spent investigating why local+DbtRunner always append --no-static-parser

Related issues

No response

Are you willing to submit a PR?

• Yes, I am willing to submit a PR!

[pankajkoti]

@komjera, thanks for filing the issue. It would indeed be nice if you could contribute this with a PR :) Looking forward to it!

[pankajkoti]

Also, could you please share some insights on how you see the static parsing benefiting you and to what extent if we didn't have that flag? We added that flag in PR #1760 as a workaround to tackle issue #1751

[komjera]

I was hoping for some parsing speedup because most of our models use standard or no macros at all, and according to dbt documentation we should see a significant speedup using the static parser.
https://docs.getdbt.com/reference/parsing#static-parser
I would estimate that 75% of our 200+ models are in scope for the static parser.

I guess I could experiment with invocation mode subprocess instead of DbtRunner, but as the documentation highlights it will probably be slower even though static parser is available in that invocation mode.

If you still agree that a disclaimer is a good idea, I can submit a PR. Do you think getting-started/execution-modes is the correct place to add the disclaimer?

[pankajkoti]

Yes, adding the disclaimer is definitely a good idea, including that this change applies since version 1.10.1 of Cosmos, and I think invocation-modes section would be a better place for it. What do you think?

We didn’t observe any degradation when using the --no-static-parser flag on our example projects; however, it’s possible that you may see an impact on your project. Could you try running the dbt CLI command locally both with and without the flag, and compare whether it makes a difference?