Improve discovery/listing of deployed processes

[fmigneault]

At least up to version 1.5.x, only 4 variants of process listing are available :

• GET /providers/{provider-id}/processes which lists all remote & visible processes of that provider
• GET /processes which lists all local & visible processes
• GET /processes?providers=true which lists all local+remote & visible processes (essentially equivalent to combining previous two requests with iteration over all providers)
• all above with added detail=false which returns only IDs instead of full descriptions

In every case with detail turned on, we can get extremely verbose responses depending on the number of deployed processes. It is very hard in this case to search of a given use case or suitable process (even more so if doing paging as described in #55). On the other hand, with detail=false, only IDs often do not provide sufficient information to know if the process is appropriate for a given task/use-case.

We should allow further requests parameters to support more advanced searches. A few ideas:

• [ ] Support keywords which are usually provided in the processes description.

  • Maybe even support , for AND and | for OR... (can't use & that separates request parameters).
  • Negation using ! or ~.
  • Automatically obtained if CQL2 supported

• [ ] Support description and title search

• [ ] #626

• [ ] #640

  • see https://docs.ogc.org/DRAFTS/21-065.html
  • see https://portal.ogc.org/files/96288)
  • see https://github.com/eseglem/pycql2 (Pydantic models for auto-schema validation and definition + JSON/Text converters) Explore other variants as needed.

• [ ] #645

• [ ] #642

• [ ] #627

  • relates to #107 (process versioning)

• [ ] #644

• [ ] #643

• [x] datetime to search for deployment/creation time (see #236 for format references)

• [ ] #641

• [ ] #646 (must be listed for all above)

[fmigneault]

@dbyrns @tomLandry Adding you guys for any docs/planing of projects and/or ideas.

[dbyrns]

We have a task in DACCS project to explore the catalog aspect of process listing. NLP should indeed be involved.

[tomLandry]

Ok understood. Aside from DACCS, I'll be on the lookout for good practices or recommendations from OGC in the testbeds. This might materialize late summer by itself, so let's keep in touch.

[fmigneault]

We have a task in DACCS project to explore the catalog aspect of process listing. NLP should indeed be involved.

This is why this entry is there 😉