Working on: Recorder Integration

I’ve been spending my time doing integration testing and defect fixing of the DDS Record and Replay application I wrote.  The Recorder records to file multiple DDS messages that are coming in at 16Hz per topic and at a total rate of about 70MB/s (ish, from the top of my head), continously.  The recorder serialises the DDS samples, puts them in a queue and then writes them out to files spread across on multiple disks (so we can handle the data rate).  The Replay functionality reads these files in, deserialises the data and sends it back out over DDS within a fine timing tolerance.  This is driven from the HCI which sends commands via DDS to start and stop the record/replay process.  The HCI was written by the guys at the other end of the M3 so that can be a bit tricky to organise and work with due to location and SVN issues.

So I’ve been doing integration and stress testing of this functionality, which has uncovered some interesting bugs, but thankfully nothing awful.  The UML model for all of this has needed updating because things changed significantly since it was originally designed.  The joys of Enterprise Architect and sequence diagrams…

I’ve spent all of today updating the DDS library I maintain to add new fields, update the test harnesses and package it for release.