Contents
The following sections provide the history of new features, updates, and resolved issued in previous TIBCO StreamBase® 10 releases.
This section provides a list of errors corrected in release 10.3.0 of TIBCO StreamBase®.
| Fixed in 10.3.0 | |
|---|---|
| Number | Resolution |
| TIBCO StreamBase® 10.3.0 incorporated all fixes resolved in the release 10.0 series through 10.3.0 and all fixes resolved in the release 7.7 series through 7.7.4. | |
| SB-44502 | If a Query Table or Table Schema was based on a named schema, then the partitioning drop-down menu in the table editing toolbar might have been disabled. It is now enabled, thereby allowing you to specify a partition when appropriate. |
| SB-44475 |
In previous releases, the querytable
administration target would only work for query tables in a top-level module.
This was fixed and query tables in nested modules are now correctly handled.
|
| SB-44457 | Specifying one or more substitution name-value pair definitions in a fragment launch configuration (as opposed to a naming a file of substitution name-value pairs), would not actually use those definitions during the launch. This was fixed. |
| SB-44202 | In previous releases, having an EventFlow fragment that contained a Query Table with no data distribution policy prevented the running a JUnit test. Now unit tests do run in these circumstances, with an INFO message issued. Tuples are added to the Query Table, but the data is not partitioned. |
| SB-44168 | An issue prevented the OSI PI adapter from loading the .NET OSI PI DLL. This was fixed for the entire OSI PI adapter suite. |
| SB-43899, SB-29388 | When logging off the StreamBase HP FIX engine, the FIX adapter was put into a state that prevented the logon command from working when attempting to log on again. This was fixed. |
| SB-43848 | An issue prevented the Kafka Producer adapter from closing on error, regardless of the option for doing being set. This was fixed. |
| SB-43797 | The Kafka Consumer and Producer adapters were fixed to address an issue that prevented suspend and resume behavior from working correctly. |
| SB-43648, SB-43458 | Artifact deployment could fail when the PMML model operator pulled in artifacts from an Artifact Management Server (AMS) via the Artifact Distribution Service (ADS) — and when the default container was explicitly set for deployment in AMS. Now, the default container name is no longer required for deployments. |
| SB-43783 | This release fixes an issue that prevented new models from completely replacing existing ones, when loaded from the JPMML Model Evaluator Operator's control port. |
| SB-43621 | There was a problem where duplicate entries could mistakenly appear in the fragments drop-down list in the Manual Input view's Output Streams and Input Streams views. This problem could occur immediately after the launch dialog for an EventFlow fragment launch disappeared, when running a second EventFlow fragment launch. This was fixed. |
| SB-43566 |
The Operator.evaluate Java API would previously
return a value of zero for data types int, long, double and Boolean if their
values were in fact null. This was fixed so that null is returned in these
cases.
|
| SB-43479 | An issue caused the JPMML Model Evaluator operator to issue NaN errors when running PMML models containing special characters and spaces in the data field names. The operator now unwraps any field name that uses escaped identifier syntax before trying to compare to any field names in the actual models. This ensures the field names will now match. The unwrapped name now also appears in the log for easier reference. |
| SB-43464 | This release fixes an issue that prevented EventFlow applications from launching if an undefined data distribution policy was defined for a query table in a unit test case. A warning is now generated. Tuples are still added to the query table in shared memory, but data is not partitioned. |
| SB-43413 | All field types were updated on the TERR operator to check for a data index range. Now, a null value is returned when trying to access an index beyond what is in the TERR object. |
| SB-43293 | This release fixes an issue in Studio's module drag and drop capability that prevented file names from converting to fully qualified names, when dropped into nested projects. |
| SB-43161 | Previously, MQTT Input adapters using the same shared connection would output messages for all topics — whether they subscribed to the topics or not. Now, each adapter outputs messages only for the topics to which they are subscribed. |
| SB-43111 | Previously, Studio would return a ResourceNotFoundException typecheck error if an EventFlow module containing a Decision Table application called for another module's Decision Table file. This was fixed. |
| SB-43032 | This release fixed a URI parameter encoding issue in the Web Reader Input adapter. |
| SB-42764 |
The WITS adapter typecheck could fail if the stringNullValue adapter configuration option was missing.
Now, the adapter no longer emits an NPE error if the WITS adapter
configuration is missing that option.
|
| SB-42746 | The Web Reader adapter's status port schema was only set during typecheck and thus was required to be set at initialization, with the runtime version of the schema. As a result, the adapter was not passing capture fields to the error output port. The Web Reader adapter's status port now works correctly with capture fields. |
| SB-42394 |
In previous releases, a collision between virtual network interfaces and the
StreamBase Runtime caused (a) extremely slow node installation times on
Windows and/or (b) when more than one fragment was run at the same time,
nodes would self-destruct, leaving no evidence in log files. Virtual network
interfaces are installed as a part of virtualization hypervisor software such
as Oracle VirtualBox, Microsoft Hyper-V, or VMware Workstation. This
collision occurred only on Windows, and was triggered by installing Docker
for Windows, which requires and enables Hyper-V.
The cause of the collision was identified and corrected, and no longer occurs with StreamBase releases 10.3.0 and later. See Programs with Incompatibilities for workarounds if you are running an earlier StreamBase 10 release. |
| SB-42250 | Model artifacts deployed from the Artifact Management Server to a StreamBase cluster via the Artifact Distribution Service (ADS) would occasionally fail to deploy on all nodes. The underlying deadlock was fixed to address this issue. |
| SB-41775 | In cases when the Kafka driver detected a TCP/IP connection drop, the Kafka Consumer adapter would emit an inaccurate warning message describing the cause. The Kafka Consumer now correctly reports the failure error when the adapter exceeds the maximum number of retries. |
| Fixed in 10.2.1 | |
|---|---|
| Number | Resolution |
| TIBCO StreamBase® 10.2.1 incorporates all fixes resolved in the release 10.0 series through 10.2.0, and all fixes resolved in the release 7.7 series through 7.7.2. | |
| SB-42876, SB-42867 | An issue caused the Bidirectional Socket adapters to incorrectly set the byte order encoding. This was fixed in all Bidirectional Socket adapters. |
| SB-42875 | The epadmin enqueue command would not roll back transactions in some situations when an error occurred. For example, data in a Query Table could be committed instead of rolled back. This was resolved. |
| SB-42839 |
In previous 10.x releases, in the Query operator's Initial Contents tab, the
Load initial rows from a resource file control
failed to consider files in the project's src/main/resources directory. This was corrected.
|
| SB-42830 | In previous 10.x releases, the epadmin enqueue stream command was not transaction deadlock-safe when reading from standard input. This would cause the command to hang if a deadlock occurred enqueuing a tuple. This was fixed. Any deadlocks are now correctly rolled back when reading from standard input. |
| SB-42828 | The Kafka sample could fail to run due to temporary files being created that could not be properly accessed. This was fixed. |
| SB-42790 | When using the Cluster Monitor on clusters where no StreamBase JVMs (EventFlow or LiveView) are discovered, the LiveView Web cards, EventFlow/LiveView Amount of Heap Memory In Use and EventFlow/LiveView Total Current Queue Depth could show an error because the corresponding tables were not created yet. The tables required to display these cards are now created in advance to prevent this error possibility. |
| SB-42750 |
The StreamBase Studio tooltip for the primaryMemberPattern object in the node.conf configuration file type was updated to reflect
its correct default value, .*
(all nodes matched by default).
|
| SB-42732 |
Release 10.2.1 fixes a problem that could occur when importing a StreamBase
schema from an XSD definition. When importing lists of referenced types, the
importer was incorrectly using the name of the
list, rather than the type of the list elements.
|
| SB-42704 | In release 10.2.0, fragment launches in Studio were configured to use only environment variables explicitly specified in the Environment tab of the launch's Run Configuration. Starting with release 10.2.1, Studio's default was reversed to match the behavior of the epadmin command, which is to inherit the system environment at node install time. See Editing EventFlow Run Configurations for further details. |
| SB-42658 |
In certain circumstances, configuring an availability zone in a configuration
file using the loadOnNodesPattern property would
fail to add partitions. This issue was resolved; if the local node matches
loadOnNodesPattern, then the partition is added
to the local node even if it is not the active or backup node.
For more information about availability zones, see Runtime Node Configuration and High Availability in Depth. |
| SB-42648 | In previous 10.x releases, running an epadmin command against a node that was not currently running incorrectly returned success. This condition was corrected to allow epadmin commands to report an error when run against a node that is expected to be running but is not. This change does not affect commands that are designed to work against an inactive node, such as epadmin node target. |
| SB-42625 | The resume container epadmin command previously reported failed commands as being successful. This was fixed and the command now correctly returns both an error message a non-zero return code. |
| SB-42515 | Release 10.2.1 now correctly displays system container information when running the epadmin display container command. |
| SB-42411 |
Release 10.2.1 fixes an issue in which two audit logs were created: an empty
file in the node directory's logs subdirectory,
and the real audit file in logs/audit.log.
|
| SB-42397 |
The Create a snapshot button in the toolbar of
the Clusters view now opens its selection dialog in the .nodes folder of the current Studio workspace. This
corrects an issue where (a) Windows users had to know the full system path to
their Studio workspace folder to get to the .nodes directory, and (b) StreamBase now displays the
previously invisible .nodes folder on the Mac.
|
| Fixed in 10.2.0 | |
|---|---|
| Number | Resolution |
| TIBCO StreamBase® 10.2.0 incorporated all fixes resolved in the release 10.0 series through 10.1.1, and all fixes resolved in the release 7.7 series through 7.7.1. | |
| SB-42604 | The StreamBase server process was incorrectly allowing timestamp-related functions in expressions that required pure functions. This was corrected. |
| SB-42338 | If a node deploy configuration is loaded during fragment deployment, it was not deactivated and removed when the fragment was terminated. This was fixed. |
| SB-41670 |
In previous releases, a decision table using BusinessEvents style expressions
required a leading zero for fractional values of type double. For example,
MyField > 0.02 was
required, and > .02 would
produce an error. This restriction was lifted.
|
| SB-41589 | The XML2tuple and tuple2XML operators can now use XSLT to transform the XML into forms more acceptable. |
| SB-41352 | A Java memory leak could occur in internal StreamBase Runtime-generated code for native methods that returned arrays. This was fixed. |
| SB-41047, SB-30970 | EventFlow fragment launch history is now remembered in fragment launch menu items. |
| SB-32231 | Intermittent compilation failures could occur when deploying EventFlow fragments. The cause was identified and corrected. |
| SB-31471 | Commented out substitution variables in configurations specified in a node deploy configuration file would syntax errors. They are now correctly ignored. |
| SB-31461 | Closing a window containing the StreamBase Command Prompt that was used to install a node would terminate the node. This was corrected. |
| SB-31282 |
The externalNativeLibraryPath parameter was
ignored in StreamBase engine configuration files when a fragment was
deployed. This was fixed.
|
| SB-31233 | Terminating a LiveView fragment before it completed initialization would cause the engine termination timer to expire, which in turn caused the node to be brought down with an error. |
| SB-31198 | A launched LiveView fragment could fail to shut down within the shutdown timer following a duplicate port failure. This was corrected. |
| SB-31040 |
Properties primaryMemberPattern and backupMemberPattern are being ignored in node deploy
configuration files. This was fixed.
|
| Fixed in 10.1.1 | |
|---|---|
| Number | Resolution |
| TIBCO StreamBase® 10.1.1 incorporates all fixes resolved in the release 10.0 series through 10.1.0, all fixes resolved in the release 7.7 series through 7.7.0, and all fixes resolved in the release 7.6 series through 7.6.7. | |
| SB-30812 | In previous releases, there was no means to specify a particular engine name when deploying EventFlow or LiveView fragments. Launch configuration dialogs now include an optional field for entering engine names. |
| Fixed in 10.1.0 | |
|---|---|
| Number | Resolution |
| StreamBase 10.1.0 incorporated all fixes resolved in the release 10.0 series through 10.0.1, and all fixes resolved in the release 7.6 series through 7.6.7. | |
| SB-30454 | Release 10.1.0 fixes an issue in which Studio was not correctly resolving cross-project resource references. |
| SB-30444 | An EventFlow typechecking error occurred when referring to adapters defined in a dependent project. This was resolved. |
| Fixed in 10.0.1 | |
|---|---|
| Number | Resolution |
| TIBCO StreamBase® 10.0.1 incorporates all fixes resolved in the release 7.6 series through 7.6.6. | |
| SB-30293 |
When you create a new project, Studio generates a test file that references
the name of the EventFlow fragment just created. In some cases, Studio did
not use the fully qualified name of the fragment, but instead used only the
basic name such as FTPGet.sbapp, instead of
com.tibco.sb.sample.ftp.FTPGet.sbapp. This was
resolved.
|
| SB-30284 | In 10.0.0, if a Java fragment was run in debug mode on Windows, selecting the button could lead to a failure where the symptoms included the appearance of a Runtime Error dialog. This problem was fixed. |
| SB-30229 | When connecting to a JDBC data source, the JDBCDataSource configuration file's userName and password parameters were ignored in some cases. This problem was resolved. |
| SB-30220 | An error caused a newly created package folder to appear twice in the Project Explorer. This was fixed. |