Use of relative paths in import statements

Hello Chris and Calin,

It would be great if we could discuss this afternoon and receive guidance on a topic that just popped up in the ETSI NFV profiles.

The guidance would determine a decision we need to take in SOL this week as it affects the publishing of the next Release-3 version (3.6.1) which is currently under remote consensus process.

As you may know, as part of SOL001, we deliver a set of YAML files containing the ETSI profile. NSD users would then import the etsi_nfv_sol001_nsd_types.yaml and VNFD users would import the etsi_nfv_sol001_vnfd_types.yaml.

These two files import in turn the etsi_nfv_sol001_common_types.yaml (among others).

So far the import statements use absolute paths. However, this is going to cause a parsing error from now on, because we are introducing a parsing check before publishing (to avoid to repeat a problem we had in the last version for not doing that). And the absolute paths are not available before publishing.

The solution that has been proposed is to use relative paths instead.

And here is the question, which has been brought up by Juan. TOSCA, clause 3.6.8.2.4, specifies for the use of relative path in import statements:

Relative path - If the <file_URI> is a relative path URL, perhaps pointing to a Service Template located in the same CSAR file, then it SHOULD cause the locally accessible Service Template to be imported.

o If the ârepositoryâ key is supplied, this could also mean relative to the repositoryâs URL in a remote file system;

o If the importing file located in a CSAR file, it should be treated as relative to the current documentâs location within a CSAR fileâs directory structure.

Strictly speaking in our case, we are not in any of the two conditions described in the bullets (we are not using the repository key and the importing file is not in a CSAR).

However, I am not sure if these two bullets are to be read as exhaustive list, or as examples. To me considering the relative path as relative to the path of the importing file is in the spirit of the definition.

Looking forward for a discussion this afternoon.

Best regards,

Arturo

tosca message