Tue 04 Sep 2007 08:32:14 AM UTC, comment #1:
I've just dive into the code for a while
and I agree this seems to be a bug.
We should certainly loop over "req->regions".
The fact is in the NetworkMessage class there is both:
I do not know this part of CERTI [yet :)]
so I need to understand the role of
vs
in order to find a precise answer and fix for this.
However this work is currently not very high in
my CERTI work stack.
Is this 'blocking' for your CERTI usage?
Do you you have a test case which exhibit a wrong behavior?
Did you try a patch on your side?
|
Sun 13 May 2007 02:35:08 PM UTC, original submission:
RTI::ObjectHandle
RTI::RTIambassador::registerObjectInstanceWithRegion(
ObjectClassHandle object_class,
const char *tag,
AttributeHandle attrs[],
Region *regions[],
ULong nb)
We can see that the "regions" parameter include multi regions, and certi use "build_region_handles" function to transform it to the RegionHandle array. But the RTIG did not process the RegionHandle array correctly:
In RTIG_processing.cc
// ----------------------------------------
// processRegisterObjectWithRegion
void
RTIG::processRegisterObjectWithRegion(Socket link, NetworkMessage req)
{
NetworkMessage rep ;
rep.object = federations.registerObjectWithRegion(req->federation,
req->federate,
req->objectClass,
req->label,
req->region,
^^^^^^^^^^^
req->handleArraySize,
req->handleArray);
......
}
As shown, RTIG only get the "region" but not the "regions".In my opinion, it's a bug.
|