Guidance about use of MAP=postgres connection string

[jemacchi]

Hi, I'm having some troubles with using MAP and a connection string to a postgres source, from which we expect to load a QGIS project which at same time contains references to layers defined in the same database.

Important:
the decoded connection string "postgres://postgres:password@postgres.database.url.com:5432?sslmode=require&dbname=postgres&schema=myschema&project=my_qgis_project_indb" points to my_qgis_project_indb project. That project was saved from QGIS to the database and layers defined on it are postgis based layers. That is those "tables" are located in the same schema. Then you have the qgis project table + all the tables/layers in the same schema

A call like this:
https://localhost:port/wps/ows/?SERVICE=WPS&VERSION=1.0.0&REQUEST=DescribeProcess&identifier=model:my_model&MAP=postgres%3A%2F%2Fpostgres%3Apassword%40postgres.database.url.com%3A5432%3Fsslmode%3Drequire%26dbname%3Dpostgres%26schema%3Dmyschema%26project%3Dmy_qgis_project_indb

should be allowing to show the list of available layers in the ows:AllowedValues tag under the input (for the model, which is simply getting a vector layer as input), but list of layers is empty.

If instead of referencing the MAP to a postgres, I use the file equivalent (that is the qgs file with same layers defined) then I can get the list of allowed values (vectorial layers).

I'm not sure if I'm missing something in middle or if it's a kind of bug or expected behavior (as the QGIS project in this case is loaded dinamically - cause it's not a known "project", as you receive it by MAP parameter - then not sure if the projects cache is considering that. The weird thing is that it's not showing an error, it just simply returns an empty list of layers)

Any thoughts/ideas ?