Running server for the first time takes too long to start for some (complex) ontologies

[dgarijo]

The implementation server we have right now takes too much time to start the API (around 10 min).

It's true that this has to be done only once when generating the server, but testing custom queries and detecting errors takes too long to debug.

Instead, we should figure out what is taking so long in the openapi generator, probably some circular references that are being validated.

Can we figure out how to reduce this time?

[mosoriob]

The implementation server we have right now takes too much time to start the API (around 10 min).

This happens with the model catalog ontologies.
So, hey new user, please don't run away from this project. For simple ontologies works fine.

[mosoriob]

This problem has been caused by the package https://github.com/p1c2u/openapi-spec-validator. I'm going to fix it.

[mosoriob]

Workaround:

1. Open the file lib/python3.8/site-packages/connexion/spec.py
2. Comment the line self._validate_spec(raw_spec)

class Specification(collections_abc.Mapping):

    def __init__(self, raw_spec):
        self._raw_spec = copy.deepcopy(raw_spec)
        self._set_defaults(raw_spec)
        #self._validate_spec(raw_spec)
        self._spec = resolve_refs(raw_spec)

[dgarijo]

Thanks. It's not a pretty solution, but we'll use it for now. We'll add an option in OBA that will comment out the line in the Dockerfile when installing the server.

[hvarg]

To disable validation on the API: https://github.com/mintproject/model-catalog-api/pull/132