REDHAWK 2.2.7
The effort in REDHAWK 2.2.7 focused on:
- Plugin infrastructure created for GPP to track custom system metrics
- Queue flush reported on all affected streams in the incoming queue
- When reading Bulk IO streams, allow the consume size to be greater than the read size (equivalent to read+skip)
- Resolving an issue with intermittent failures in
omniEvents
when writing tosyslog
running on CentOS 7. - Components generate from command line with applications like createBinaryComponent generate the correct header on the SCD and PRF files
- Java message ports are consistent with Python and C++ message ports. When the message fails, it tries again in subsequent messages
- Python helper pythonComplexListToBulkioComplex converts numpy.float to Python native float
Fixes
The following fixes were completed in REDHAWK 2.2.7:
Core Framework Fixes
- Resolved an issue with device allocations. If the device's state is
BUSY
, the device allocation always returnsFalse
. - Resolved an issue with ports generated from custom IDL. The ports now use the hierarchical logger to log errors.
- RH_SOFTPKG_CXX version checks are honored when minimum and maximum versions are given for the dependency
- Resolved an issue with custom interfaces returning a data structure with variable-length fields. The generated code now returns the correct type.
- Resolved an issue with the DeviceManager. It now sets the process group id for all services and devices that it launches.
- Resolved issue with attachable ports (like dataSDDS and dataVITA49) failing silently. They now report a failure to attach with an exception.
- Resolved an issue with code generators. They no longer fail when the
wavedev
file is missing. - Appropriate XML headers are created for SCD and PRF files when creating a project from command line
- Python helper pythonComplexListToBulkioComplex converts numpy.float to Python native float
- When invoking resource.getPOA, if a name is provided, the named POA is returned
- When using Bulk IO streams, the amount of data consumed (pointer pushed forward) can exceed the amount of data read
- Java output message ports can recover from CORBA.COMM_FAILURE errors
Documentation Fixes
- Updated
custom.css
to fix table formatting issue. - Removed Standalone EPEL information from the REDHAWK External Dependencies appendix.
Special Resolution
The following previously reported Known Issues have been closed in REDHAWK 2.2.7 with the following resolutions:
Core Framework Special Resolutions
popen
inspawnnodebooter
needs anb.wait()
to make sure that a return code is received - Cannot Reproduce - This bug could not be reproduced against REDHAWK 2.2.6. (The test framework already has a wait on thepopen
.)- Out scan control ports are missing from FEI library - Duplicate of Output scanning tuner ports are missing the template, and this feature was resolved in REDHAWK 2.2.6.
- Sandbox generates an unnecessary error when cleaning up - Fixed by Other Issue - Fixed by update in REDHAWK 2.2.3 release: When the Python sandbox exits and it has a reference to a Device Manager, the deactivation of the object no longer raises an exception if the object no longer exists.
- Cannot call
connectPort
on a bidirectionalMessageEvent
port - Won't Fix - This is the expected behavior. A bidirectional message port is a message consumer port. Connecting a message consumer port to another message consumer port is invalid.