GitLab

The later has been skipped, while it makes sense sometime to specify it during
profile definition.

The former are skipped to make the UI for Attachment or URI on BinaryDataObject
much clearer.

Closes #13751394 and #13830622

Closes #13853636

Flow is simpler when using empty tuple instead of empty string.

Use 'seda-2' as language code to please rdflib which check language code validity
using a regexp. To do so, also extend skos cube limit size of 5 on language code.

Closes #13008700

[ui] Display message instead of 'search' button when we're missing information to resolve constraint

Closes #13853304

Closes #13853501

and drop related monkey-patch.

And not globally.

Related to #13830428.

Since this is mandatory in data model.

In that case, we shouldn't check for existence of an entity under the choice

A content has no more than one title and description. In the same way, an
ArchiveTransfer has no more than one comment.

Handle this as hardcoded special cases, waiting for a better solution.

Closes #13750725.

Iterate over all etypes and rdefs constituting the compound graph of a
`SEDAArchiveTransfer` and set the same permissions than the
`SEDAArchiveTransfer`.

Instead of displaying the DataObjectReference entity. And put this information
in the `relations` section.

Related to #13751449.

Closes #13830646, #13853099.

Closes #13813234.

On archive unit form, hiding the choice field was handled wrongly by editing
xsd2uicfg.py The real problem is we need to load uicfg rules before the views.

As I can't make it appear on the inlined form title, make the overall message
look better by omitting obscur relation name.

Also, change translation to use alternative instead of xsd:choice, as in the UI.

This code wasn't tested and nothing was relying upon it (those elements are in
inlined form in the UI).

And hide them from the default tab