[sdk/python] Adds a default exception when dependency cycles are created

[kpitzen]

Description

Currently, when we detect that we've created a cycle in the dependency graph, we early-exit. This works well enough for simple cycles, but when early exiting is not sufficient (as when providing a resource output as an argument to another resource's inputs), we will still fail to resolve the full dependency graph. This PR introduces an exception-by-default, as any cycles represent an unsafe/invalid dependency graph and should be resolved manually. We also provide an escape hatch to fall back to current behavior, in case users would prefer to retain the ability to create unsafe dependency graphs (potentially introducing infinite hangs when resolving those graphs).

Fixes #13551

Checklist

• I have run make tidy to update any new dependencies
• I have run make lint to verify my code passes the lint check
  • I have formatted my code using gofumpt

• I have added tests that prove my fix is effective or that my feature works

• I have run make changelog and committed the changelog/pending/<file> documenting my change

• Yes, there are changes in this PR that warrants bumping the Pulumi Cloud API version

[pulumi-bot]

Changelog

[uncommitted] (2023-11-17)

Bug Fixes

• [sdk/python] Introduces RuntimeError when we detect a cycle upon adding dependencies to the graph. Additionally adds "PULUMI_ERROR_ON_DEPENDENCY_CYCLES" as a new environment variable to control this behavior. Set to False to return to the previous behavior, which could potentially re-introduce infinite hangs for some programs.
  #14597

[kpitzen]

Additional comment - what's the best way to document this new python-specific env var? Being a bit more explicit in the changelog? Not sure what the process looks like now.

[Frassle]

Additional comment - what's the best way to document this new python-specific env var? Being a bit more explicit in the changelog? Not sure what the process looks like now.

Firstly it should be prefixed with "PULUMI_", secondly yes probably changelog, but also you can add it to the sdk/env file even if none of the Go code currently uses it yet. Will then show up in pulumi env. I'd guess if we need to do similar fixes in other SDKs we'd then make sure we use the same envvar.

[kpitzen]

Additional comment - what's the best way to document this new python-specific env var? Being a bit more explicit in the changelog? Not sure what the process looks like now.

Firstly it should be prefixed with "PULUMI_", secondly yes probably changelog, but also you can add it to the sdk/env file even if none of the Go code currently uses it yet. Will then show up in pulumi env. I'd guess if we need to do similar fixes in other SDKs we'd then make sure we use the same envvar.

Got it - it is prefixed with PULUMI_ already, but good call on adding it to the go env file. I'll do that.