Specifications

9032277-04 Software Superseded by CS3/MMS3
5-31
Patches Issued After the CS1/MMS1 Release
For epapi, the port to NT introduced the "sigaction_thread()" function,
which runs on its own native NT thread upon call to EPI_fcntl(). This
function is implemented to always synchronize with the
Cs_EPI_pause() method, to be called from the application thread.
Unfortunately, this implementation changes the EPI interface, now
requiring that Cs_EPI_pause() be called in order for ANY IO
notification to occur. There is unnecessary code introduced by the port
which actually causes a segv for large messages. This has been
resolved by removing the EPIAcceptEvent from sigaction_thread and
Cs_EPI_pause. This will allow true async IO notification from epi.
This has also been resolved by removing the unnecessary code that
introduced the segv for large messages.
Support has been added for the Catalyst wsx5201r and wsx5225r
boards.
After upgrading from 4.0rev3 to 5.0rev0, you can no longer view or
convert 40rev3 statistics with the 5.0rev0 version of ddm_load. After
doing the upgrade, if you use ddm_load to reload the 4.0rev3 StatDb
from a ddmsavefile then use the SDE application, the common
attributes only show the "0x0" for available data. Using the Reports
application you will see an available range for data prior to the
installation of 5.0rev0, yet when you run a statistical report for a
router the report indicates "Completed Successfully - No Records
Found". The problem is that the 4.0rev3 DDM saves do not include
Attribute names or types. Because of this, we don't create the AttrDict
records in the new DDM database, as part of the migration load. This
has been resolved by writing a tool that can be run after migrating
from 4.rev3 to 5.0rev1, that will update the DDM database with all
the relevant Attribute data from a running SpectroSERVER.
If an application sets the instance name member of the CsDataList
object, the memory never gets reclaimed. This results in a memory
leak. This has been resolved by writing a destructor to reclaim the
memory.
Connection Manager in the Alarm Manager reports a DOWN
connection for the View Service, incorrectly. This inhibits the ability to
navigate to device from the Alarm Manager when need be, because the
Connection Manager is reporting a DOWN connection. The resolution
to this problem does not change the view service advertisement when
a SpectroGRAPH is brought down with one already running, but
prevents map storms from freezing up SPECTRUM's processes.