The table must be recreated if the target column does not exist.
Also temporary column name should use double underscore as module may use
single underscore.
issue7978
review52591002
- remove invalid escape sequence
- use read_file instead of read_fp
- use importlib instead of imp and __import__
- use callable instead of instance of collections.Callable
- fix silent file closing
- replace SafeConfigParser by ConfigParser
- replace decodestring by decodebytes
- replace encodestring by encodebytes
issue7666
issue7390
review53361004
This prevents to keep the minimal connection on database that are not used
anymore. And set minimal connection to 0 for template1
issue7531
review62261002
- When recreating the table to drop a column, the correct SQL type must be
converted by the database.
- SQLite uses BOOLEAN as base type for BOOL
issue7120
review37001002
If pg_dump/pg_restore are not available, we create a copy of the database on
PostgreSQL server using the template option of "CREATE DATABASE". We still
create a file in the cache folder to know that there is a backup.
issue6699
Some back-ends have constraint on the length of data name, they usually
silently truncate them, but this can create collision. The solution is to hash
the name when it is too long.
issue4620
issue6494
review37381003
There is no guarantee that the method will not raise a database error on the
first registration because it can rely on other tables being registered.
So the method must be called only when needed. This is when the table already
exists and has records. In this case other tables also already exist.
issue6409
This changeset defines on the field the (somehow) abstract SQL type a field
should use letting the duty of specifying the real SQL type the database
supports and the conversion to/from the python types to the backend.
issue6351
review33951002
As now the cache is managed by the Transaction, when creating the database for
the tests, the ir_cache table should be present before the registration of the
ir module. Also the transactions to create or drop the database should not have
cache because either the database is not created at the start or dropped at the
commit.
issue6216
review29171003
It is too complex and expensive to lazy create the connection pool because all
methods which uses the pool needs to acquire a lock. So it is easier to create
it when the instance is also created. In this case, on closing the database, it
should be unregistered from the instances dictionary.
issue6202
review32991002
Cédric Krier