[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[xmlblaster-devel] [Fwd: Re: [xmlblaster] Ldbc]

Hi Peter,
I moved this topic to the developer list since it is more appropriate.
I fixed the property parser to allow commas inside subproperties (for the mapping). You just need to enclose the whole subpropery with double quotes.

Tomorrow I would add a flag "cascadeDelete=no" (yes per default) and then we should have it all to be merged into JdbcQueueCommonTable, making configuration, documentation fixes and testing simpler to be done. (for the same reason we removed a while ago the original JdbcQueue).

Cascade Deleting: I have not looked at it.


Michele Laghi wrote:
Hi Peter,
I have looked at the code and in fact it should be possible to implement all in one single common code by adding two more configuration parameters.

Paring the mapping should be possible after some code modification (the ',' inside a mapping should not break in future)

Before I start however I want to make sure the current code works. I tested the ldbc with postgres and there is one only error in the testsuite. It fails when making the request in peekBySamePriority when 12 entries in the queue and 4 entries expected to be peeked. The statement:

"SELECT * from TEST_ENTRIES where queueName='callback_QueuePluginpeekMsg' and nodeId='xmlBlaster' and prio=(select max(prio) from TEST_ENTRIES where queueName='callback_QueuePluginpeekMsg' AND nodeId='xmlBlaster') ORDER BY dataId ASC"

with the error

[Nov 18, 2003 10:05:08 PM TRACE LdbcPreparedQuery] Constructor. SQLException: 37000

Any idea about the cryptic 37000 ?

About the tests: I think it is not necessary to write new tests: the criteria that the old tests should pass should be sufficient.


Hmmm strange... Postgres was the only db I was fully confident would work... I will look into the error... They are not that easy to track down. Thanks for the test.

As I said I am not sure I am erasing enough from the tables with the
replacement code to overcome no cascading delete in ldbc however that
may not explain the test results. Surely there are other tests that
erase and check.

As for your other suggestion about integrating the code... I have no
problem with this however we should wait for a while and get it stable
and working first.


Michele Laghi
mailto:laghi at swissinfo.org
tel. +46 8 7492952 / mob. +46 70 4103964