The problem of a unique human-readable id vs UUID

[fguerra06]

Hello,

I'm dealing with a problem of articulation between UUIDs and human-readable identifiers.

Sorry for the long message but I'll try to be clear...

I have two postgis tables: « polygon » and « points ». Points can be children of polygons.

In Postgis as in Qfieldsynch, the primary key of both tables are uuids. Here is an example with the polygon table:

Each polygon has a type and a number, the combination of them must be unique. This is my auto-calculated and human readable “type_num” field.

In the “points” table, the “fk_polyg” field is a relation reference widget. The display expression is the “type_num” field in the polygon table.

The “id_pol_pt” field is auto-calculated using the “represent value” function. It is my unique human-readable point identifier. It concatenates “fk_polyg” which is supposed to be the display expression “type_num”, the type of point and the point number. « type_num » is itself the result of automatic concatenation in polygon table.

Point labels and display names are based on the « id_pol_pt » field.

When I add a point directly in Qgis, everything works fine. I can see my human-readable identifier.

After exportation, this identifier is recognized by Qfield and all goes well when I add a new point in Qfield. I can see my human-readable field.

The problem occurs when Qfield and Qgis synchronize. QGIS is unable to find the “type_num” component of the polygon. It writes the uuid instead.

I get the following error about the RepresentValue() fonction.

"2025-02-06T15:14:23 WARNING Layer point, field fk_polyg: representValue() with inconsistent layer parameter w.r.t relation referencingLayer"

Is there something wrong with my approach?

Thank you.

Fabien