This does not change code flow:
- self.master_node is None when entering the loop for the first time
- there is no way to restart the loop other than a PrimaryFailure exception

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1686 71dcc9de-d417-0410-9af5-da40c76e7ee4

This is identical to what is done in master/app.py:run.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1685 71dcc9de-d417-0410-9af5-da40c76e7ee4

This reduces the cost of invoking log method when the log line is not to
be emitted.
Also, don't put a space before a colon.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1684 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1683 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1682 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1681 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1680 71dcc9de-d417-0410-9af5-da40c76e7ee4

This bug was introduced in r928 and causes caches invalidations to be
effectively no-ops.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1679 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1678 71dcc9de-d417-0410-9af5-da40c76e7ee4

Ask/Answer packets are linked together in both direction.
The response flag is automaticaly assigned on the answer packet.
The special 'Error' packet can be used as an expected answer for any
non-ask packet.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1677 71dcc9de-d417-0410-9af5-da40c76e7ee4

A disconnection case was calling the handler method before the close(), fix
it and factorise connection closure.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1676 71dcc9de-d417-0410-9af5-da40c76e7ee4

The NotifyNodeInformation packet was not always sent before connection
closure, call abort() to ensure the write buffer is flushed before closure
so the peer can understand that it is not a primary failure.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1675 71dcc9de-d417-0410-9af5-da40c76e7ee4

Aborted connections are supposed to be closed but remains until the write
buffer is sent. Those connections must not be used for common operation
(broacast node information...).

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1674 71dcc9de-d417-0410-9af5-da40c76e7ee4

Allow override the defaults to run unit, functional or both test suites.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1673 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1672 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1671 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1670 71dcc9de-d417-0410-9af5-da40c76e7ee4

Add a related test case.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1669 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1668 71dcc9de-d417-0410-9af5-da40c76e7ee4

Keep it enabled by default so tests run with it.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1667 71dcc9de-d417-0410-9af5-da40c76e7ee4

This happens from time to time in tests, most probably during shutdown as
it doesn't make tests fail. This will help tracking this bug down.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1666 71dcc9de-d417-0410-9af5-da40c76e7ee4

This allows automating tasks right before test fails (for example, dumping
database state) for later inspection.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1665 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1664 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1663 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1662 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1661 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1660 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1659 71dcc9de-d417-0410-9af5-da40c76e7ee4

- FinishTransaction -> AskFinishTransaction
- LockInformation -> AskLockInformation

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1658 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1657 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1656 71dcc9de-d417-0410-9af5-da40c76e7ee4

Either the transaction got the load lock for all its objects, or it got it
for none. So instead of having a "try..except" block which either passes
or catches for each iteration, use a simple test.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1655 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1654 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1653 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1652 71dcc9de-d417-0410-9af5-da40c76e7ee4

Just being known to the same transaction doesn't propagate isolation
from one connection to another.

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1651 71dcc9de-d417-0410-9af5-da40c76e7ee4

This uncovered the bug fixed in r1553, and also uncovered a bug in
test_01_EpollEventManager (fixed in this commit).

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1650 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1649 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1648 71dcc9de-d417-0410-9af5-da40c76e7ee4

git-svn-id: https://svn.erp5.org/repos/neo/trunk@1647 71dcc9de-d417-0410-9af5-da40c76e7ee4