HPCC-15582 Revert HPCC-15489 and give slave managers own cache

[jakesmith]

Signed-off-by: Jake Smith jake.smith@lexisnexis.com

[hpcc-jirabot]

https://track.hpccsystems.com/browse/HPCC-15582
Jira updated

[jakesmith]

@ghalliday - please review

[HPCCSmoketest]

Automated Smoketest
Sha: f8bb024
Build: success
Install hpccsystems-platform-community_6.0.0-rc5.el6.x86_64.rpm
HPCC Start: OK
HPCC Stop: OK
HPCC Uninstall: OK

[ghalliday]

minor: normally activityId wouldn't be a reference parameter

[ghalliday]

@jakesmith a few minor comments.

I think the change will work. I'm not a great fan of passing in the array of slave allocators, but it is relatively clean, so no big objections.

[jakesmith]

I'm not a great fan of passing in the array of slave allocators,

An alternative I supposed would be to either have a :

globalRowManger->addSlaveRowManager(<cache>)

which create the slaveManager with provided cache at the time.

Or a :

globalRowManager->setSlaveRowManagerCache(unsigned slaveId, <cache>);

.. i.e. used default globalRowManager cache, unless this was called to override.

@ghalliday

[HPCCSmoketest]

Automated Smoketest
Sha: 2ed51b3
Build: success
Install hpccsystems-platform-community_6.0.0-rc5.el6.x86_64.rpm
HPCC Start: OK
HPCC Stop: OK
HPCC Uninstall: OK

[ghalliday]

I'll merge as-is. We can possibly think about restructuring the parameters at a later date.