Hi Francesco, > Gesendet: Dienstag, 10. Dezember 2013 um 10:20 Uhr > Von: "Francesco Chicchiriccò" <ilgro...@apache.org> > >> Gesendet: Montag, 09. Dezember 2013 um 22:22 Uhr > >> Von: "Oliver Wulff" <owu...@talend.com> > >> > >> @Guido, is the property "connectionProperties" sufficient which is a > >> semi-colon separated string: > >> http://commons.apache.org/proper/commons-dbcp/configuration.html > > No, connectionProperties only sets properties on the JDBC driver, not on > > the connection pool. > > > > If the persistence.properties file is considered to be too polluted with > > too many connection pool > > settings, I see two possibilities: > > > > * only make the most important ones configurable externally (I've suggested > > some below); > > people who want to change more settings can still overwrite > > persistenceContext.xml; and/or > > * use the Spring default mechanism for properties > > (i.e.${jpa.pool.maxActive:8}); thus the properties that keep their default > > value can be omitted in the properties file, with the disadvantage that one > > does not immediately see that they are configurable, and the advantage that > > there is no need to change the configuration file of existing Syncope > > installations. I'd would be even better if one could make Spring not set > > the value at all (i.e. use DBCP's default) if not specified, but I don't > > think this is possible. > > > > WDYT? > > I'd say latter option (e.g. Spring properties with default values) is > fine: would you mind to grab Oliver's patch, apply your changes and > re-attach?
Sounds good. I can do this, but cannot promise exactly when (currently I'm very busy with other things unfortunately). > > @Francesco: I'd be interested in why ISOLATION_READ_COMMITTED was chosen to > > be set explicitly as a transaction level (but only on localDataSource, not > > jndi) - was there a specific reason? > > A quick search of mail archives (via > http://syncope.markmail.org[http://syncope.markmail.org]) gave > this commit (released as part of 1.0.2-incubating) > > http://svn.apache.org/viewvc?view=revision&revision=1392392[http://svn.apache.org/viewvc?view=revision&revision=1392392] > > which is bound to > > https://issues.apache.org/jira/browse/SYNCOPE-124[https://issues.apache.org/jira/browse/SYNCOPE-124] > https://issues.apache.org/jira/browse/SYNCOPE-202 > > SYNCOPE-202 in particular refers to integration tests problems with > MySQL (and Oracle): here's why the isolation level was set to the > DataSource that was supposed - till this discussion - to be only used > for integration tests. > > This fact also leads to the consideration that before committing any > patch, all supported DBMS needs to be checked. Thanks, that is very interesting. If changing the isolation level can lead to potential problems, I think it would be a good idea to deal with this separately (i.e. restricting SYNCOPE-460 / the corresponding patch to the DBCP configuration and raising a separate issue for possibly making the isolation level configurable). Cheers, Guido
