Currently, we foist the original config from the job into execute_in_process. This means that we might end up with executor config that doesn't make any sense. This diff makes execute_in_process ignore that extra config.
Details
Details
Added test to ensure executor config is ingored
Diff Detail
Diff Detail
- Repository
- R1 dagster
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
| python_modules/dagster/dagster/core/definitions/run_config.py | ||
|---|---|---|
| 98 | This is sus, but I think it's just that pylint can't infer execute_in_process_executor is an ExecutorDefinition. | |
| python_modules/dagster/dagster/core/definitions/executor.py | ||
|---|---|---|
| 235 | maybe expand that it triggers special behavior in the config system | |
| python_modules/dagster/dagster/core/system_config/objects.py | ||
| 191–203 | why do we need this? Is it solvable by changing the config schema on the in process executor? if not add a clear comment here | |
| python_modules/dagster/dagster/core/system_config/objects.py | ||
|---|---|---|
| 191–203 | We might be able to get around it, but I think a config schema change would be more confusing. Essentially I think we would have to change the name of the executor to be whatever the name of the expected executor is, which would be a rough experience, and harder to explain than this IMO. | |