cubicweb #1089207 do not encode Unicode strings from RQL statements to db encoding [resolved]

The choice of encoding or not Unicode strings to UTF-8 should be left to the DB backend.

The current implementation will always encode unicode values to the db encoding (and if no encoding is provided in the source, UTF-8 is used by default). This works with postgresql + psycopg2 but not with SQL Server in which Unicode is handled natively, so we have to decode the utf-8 in logilab.database.sqlserver to pass unicode objects to the backend.

Additionnally, it might be worth checking if the recent versions of pg still require encoded strings.

prioritynormal
typeenhancement
done in3.9.0
load0.100
load left0.000
closed by<not specified>