Luciano Resende wrote:
On Tue, Sep 8, 2009 at 9:47 AM, Simon Nash<[email protected]> wrote:
Luciano Resende wrote:
Could you guys please try to increase the SLEEP_INTERVAL from 400L to
2000L in the Constants class [1] inside the gdata-runtime module and
let us know if this make any changes on the failures.
[1]
https://svn.apache.org/repos/asf/tuscany/branches/sca-java-1.5.1/modules/binding-gdata-runtime/src/test/java/org/apache/tuscany/sca/binding/gdata/Constants.java
I tried this but the test still failed. I then increased it to
3000L and the tests passed but took much longer to run, with
frequent lengthy pauses.
It them seems that the problem is the turnaround time on the Google
Infrastructure to process the post and make the new entry available.
I'm not sure what to do here, maybe we could increase the delay to
3000L only on the more time consuming test, or the worst case disable
some of them...
Thoughts ?
Is there any way for the client to get back an acknowledgement from
the Google infrastructure that the new entry is available, so that
it can continue safely when it gets this acknowledgement? The current
API seems to return an Entry object that may or may not be safe to
use at the time it is returned.
Simon
