cubicweb #1056023 be able to specify the schema in which DB tables are created and used [done]

For the backends which support it, having some control over the DB Schema used to store the tables can be very useful. A typical case is when the user creating the tables is not the same one as the one making queries. On some rdbms (Oracle, SQLServer are in that case, not sure for PostgreSQL), the user has a default schema linked to his login, and will not be able to use objects outside that schema unless the schema is made explicit in the queries.

prioritynormal
typeenhancement
done in<not specified>
load1.000
load left0.000
closed by<not specified>

Workflow history

from state (1)to statecommentdateUser
opendone

db-schema configuration option added in 3.20

2015/01/27 08:59 UTCjcristau

Comments

  • 2011/03/28 06:35, written by adimascio

    For postgresql, a simple "SET SEARCH_PATH TO myschema" early enough would do the trick. Otherwise, modifying cubicweb.server.sqlutils.SQL_PREFIX to be myschema.cw_ should work too.

    That said, a somewhat more general solution might be needed : a client might have specific database exploitation rules and specific table naming conventions.

  • 2011/03/28 06:36, written by adimascio

    Don't forget cubicweb-ctl db-restore / db-copy / db-dump implementations.

  • 2011/03/28 07:58

    and don't forget FTI indexes : we don't want pollution in there

  • 2011/03/28 08:03, written by adimascio

    We might want to consider tablespaces also. DBAs might want to write indexes and data on separate disks.

add comment