Grouphug - Clustering with an evil twist!
Released 28/06/2008

This is a snapshot release.
Iteration 100 test summary: 100 iterations, 2600 passed, 0 failed, 600 test groups, 2600 subtests
All tests pass! All the time!... On my machine.

CLEAR ghhome/write. Because of some fixes mentioned below, you must remove the write dictory and create a blank one. Or simply start with a vanilla copy of grouphug as packaged. The most noteable reason is the getnewid changes mean that the IDs will all reset in this release anyway since they won't have the new marker which signifies that they have already been created.

You no longer need to run the extra hacks before running the tests. It can simply be called:

 gh test.all
 gh test.all.repeat 100

Improved getnewid to be more resilient. So now it's resilient and reliable.

Increased wait time for services to start during testCaseHappydistributedQueItem. This solved the final intermittency of testCaseHappydistributedQueItem.numQueProcDuring.
It should be noted that this was an intermittency of the test and not the thing being tested, since it would start working properly as soon as it started up (hense tests later on were fine).

This is my first priority for the next release since this amplified some of the issues I was having with the event ques.

Get it.
In this release
'Proof of concept'.

© Copyright Kevin Sandom 2008 Logo