Closed denisri closed 2 years ago
ModuleConfiguration.is_valid_config()
could return additional requirements instead of just True
/False
For instance SPMConfiguration.is_valid_config()
could return {'matlab': {'mcr': True}}
or {'matlab': {'mcr': False}}
depending on the SPM standalone
state, which would add both dependency and constraints on the matlab modules (in the same way/syntax as Process.requirements
).
Those added requirements would then be checked (again?) to see if they match.
Config modules dependencies are not checked properly for execution: a config module can declare it depends on another, for instance spm depends on matlab, but the validity of this dependency is not checked and cannot be checked at execution time, and dependent modules are dropped from the execution context config.
is_valid_config()
method only receives the spm module config, and cannot check if the matlab module is configured appropriately (for standalone or matlab execution, typically).standalone=False
option, cannot require matlab with a validdirectory
, or SPM withstandalone=False
, cannot require matlab with a validmcr_directory
.The current API does not allow to configure things correctly, so it must be reworked.