Grouphug - Clustering with an evil twist!
Released 15/06/2008

This is a snapshot release, although it should be pretty together.

This release adds convienience for installation and monitoring.

testCaseHappydistributedQueItem.numprocPost was picking up the monitor, so would fail when ever the monitor is running. The monitor has been added to the list of exclusions. These changes have been applied as relevant to other tests. On initial observations testCaseHappydistributedQueItem.areWeAWorker and testCaseHappydistributedQueItem.numQueProcDuring may now be happy. It seems unlikely that areWeAWorker would have been affected by this, but I will keep an eye on it.

IDs are currently occasionally resetting when Grouphug starts and causes testCaseHappydistributedQueItem.eventCompleted 1 and 2 to fail. This is a problem with the events and ques since they rely on the IDs to know where to start reading. I hope to fix this in the alpha release. In the mean time you can get around it by running the hack ids function like this:

 gh dev.hack.ids 

Added the ability to list all functions in all files. See gh help dev dev.lsfunc

Added the ability to create a new plugin or sys module from a template. See gh help dev

Note that this version is not a service/daemon. It runs as a user. This will change for the beta release.

Added the race sys module. This handles race conditions. This could add a few dependancy bugs, but hopefully I'll have this sorted before release. If you do find any errors about ./sys/race/bin/race, please let me know. See gh help race

The race sys module is currently loaded on every execution. This is inefficient. Dependancies should be derived and move the loading to parms.

Moved dev documentation to devdoc to remove conflict with the help for the dev module.

Added mon.p which monitors grouphug processes in real-time.

Get it.
In this release
'Proof of concept'. This is make a few things more convienient in testing. Some broken tests are fixed.

© Copyright Kevin Sandom 2008 Logo