These release notes are for PhixFlow 6.0.0.
This release is for Oracle, SQL Server and MySQL.
Contents
Installation / Upgrade
To install a new PhixFlow instance, follow the standard installation instructions.
To upgrade an existing PhixFlow instance, follow the Standard Upgrade Instructions plus the special instructions for all intermediate releases.
This release has no special instructions.
Prerequisites
This PhixFlow release requires:
- Java 7
- Apache Tomcat 7
One of:
- Oracle 11g or 12c (preferred)
- SQL Server 2008R2 or later
- MariaDB (MySQL) 10.0.14 or later
Features / Improvements
Installation
The Java installer has been extended to create and populate the database tables, and to set the customer name (needed to match the licence key which is installed through the graphical user interface).
It is still necessary to use sqlplus / sqlcmd / mysql (or other tools) to create the database or schema.
See Installation and Upgrade for details.
Actions
Stream Action Lists will show the Action full name (with a parent Stream name included) when they come from different Streams.
The 'Result Includes Warnings' and 'Result Includes Error' check boxes are now ticked by default while creating a new Stream Action.
Forms
Fields on user-configured forms can now be explicitly set up as multi-line or single line.
Pressing the Enter key when entering data into a multi-line field inserts a carriage return character.
Pressing the Enter key when entering data into a single-line field is equivalent to pressing the OK button.
Previously all fields on user-configured forms were multiline.
Views
Views can now contain local attributes.
Pipes
Lookup Pipe Result Caching
Lookup pipes cache their results read in memory in order to reduce the number of database reads.
These cached results are subject to a limit on the number of records cached, but in previous releases
- this limit was only applied to previous cached results, never to the result just read, and
- if the pipe Cache Size was blank, no limit was applied.
This could result in reading too many records into memory, resulting in excessive garbage collection and loss of service.
At this release,
- there is always an effective pipe Cache Size; if the Cache Size on the pipe is blank, the System Configuration Default Pipe Cache Size is used.
- The pipe Cache Size limit is applied as the result records are being read into memory.
- There is an overall Maximum Pipe Cache Size in System Configuration; setting the pipe Cache Size greater than the maximum will cause the lookup to fail with an error.
- There is an Enforce Pipe Cache Size check box in System Configuration. If checked, pipes that exceed their cache sizes will fail with errors; if not, they will generate warnings.
If you have migrated from an earlier version of PhixFlow, Enforce Pipe Cache Size is set to unchecked so that incorrect pipe Cache Size settings do not cause existing models to fail.
We recommend that you correct the pipe Cache Size settings using the generated progress and warning messages as a guide and set Enforce Pipe Cache Size to checked at the earliest opportunity.
Directed Pipe Strategy
The constraints on the use of the Directed Pipe Strategy have been relaxed.
It is no longer necessary for a Merge Stream to have at least one Mandatory Non-Directed Pipe. Instead, a Merge Stream that has any Directed pipes must have at least one non-Directed pipe.
Note that the Mandatory flag on a Directed pipe has no effect.
Connectivity
Database Exporters
Cancelling a database export now stops the export rather than allwing it to run to completion.
It is now possible to set a limit on the number or error messages that a Database Exporter can generate. See System Configuration for details.
Email Accounts
An inbound Email Account can be configured to delete messages from the mailbox after they are downloaded. Messages can be deleted immediately or after selected number of days.
File Collectors
File Collectors now respond more quickly to Task Cancellation.
Scripting
"_outputMultiplier" has been added as the supported and recommended way to refer to attributes created by setting an output multiplier (instead of "_type", which still can be used).
System Administration
Diagnostics
Changes to the server logging settings in log4j.properties now take effect automatically (after a few seconds) without having to restart the Tomcat server.
Processing Statistics
PhixFlow generates processing statstics which are saved to the database. These statics are for use by administrators and designers who are tuning for performance.
The following stats are recorded:
Activity Stats for Taskplans started and actions triggered and also stream write.
See Processing Statistics for more information.
Reset Configuration
In order to reset the system configuration, you must now
- Open the System Configuration form (Advanced tab) and click on Allow Reset Configuration, then (within 5 minutes)
- Click Admin->Reset Configuration.
See Reset Configuration for more information.
Performance
Improved performance while doing analysis which generates lot of log messages (>1000).
Bug Fixes
Graphical User Interface
Fixed: using the date picker form to select the 31st day of the month actually selected the 1st day of the month.
Various minor fixes / improvements.
Views
Aggregate views with only one attribute (e.g. summation of an amount on an attribute) now work for PhixFlow instances installed on SQL Server.
It was previously possible to attempt to 'Edit' a Stream's Default View, even though this Edit generally failed because the View doesn't actually exist in the database.
Streams
Values saved to 'Date' type stream attributes now have their time parts set to 00:00:00.
When creating a Stream directly from a File Collector, missing Attribute sizes are set to suitable default values.
Publishing a stream failed on SQL Server if the stream table existed but had no primary key constraint.
Publishing on MySQL always failed.
Analysis
Re-booting Tomcat during Analysis, the re-running Analysis to recover sometimes caused a pipe to read the wrong data.
Collectors & Exporters
Database Collectors & Exporters
SQL Comments now only apply to the current line the comment is on, not to the remained of SQL after the comment line.
If an Excel File Collector range expression is invalid, the file collector now reports the error and stops; previously the failed range expression was ignored and data was collected as though no range expression had been set at all.
Excel File Collector : improved handling of error conditions and exception values.
Fixed: Exporting the data visible in a Stream View caused all stream sets in the Stream to be exported.
Internal variables _worksheet and _range, which are set when reading from an Excel File Collector, are now correctly validated in Stream Attribute expressions.
Email Accounts
The Email Account Test button didn't work for imap/imaps accounts.
Scripting
Functions
listToString
The interpretation of 'ignoreNullsFlag' was reversed. 'ignoreNullsFlag' now behaves as documented: 0 means treat embedded null values as empty strings, 1 means treat embedded nulls as though they were not in the list at all.
dateAdd, dateSet, getDayOfWeek, getDayOfYear, getMonth, getYear, timeBucket, timeSlots
Improved handling of null values.
Logging
Cancelled tasks are no longer logged as errors to the server log file.