Page 1 of 1
Saving the rule file and recalculation of feeders
Posted: Sun Dec 07, 2014 12:19 pm
by mvaspal
Hi
I am wondering this is a new feature in 10.2.x, I saw it in 10.2.2 now, or did I miss it in earlier versions?
I changed a rule and saved it but did not change the feeder section. The rule was saved in 1 sec, no feeder recalc was kicked off. Then I amended the feeder section, saved the rule, it took much longer because it recalculated the feeders.
If it is a new feature it is really useful because if you need to amend a rule in a cube with heavy feeders but you do not need to change the feeders then it takes a second to save the rule, enabling such changes even during peak usage periods.
Thanks
Matyas
Re: Saving the rule file and recalculation of feeders
Posted: Sun Dec 07, 2014 1:50 pm
by lotsaram
Yes. This is new in 10.2 and (generally) very useful. If feeder definition is not changed then rule save is limited to rule recompilation and feeder generation only if feeder section is changed.
Re: Saving the rule file and recalculation of feeders
Posted: Sun Dec 07, 2014 2:06 pm
by mvaspal
Hi
Thanks for quick response, I find it also great
Re: Saving the rule file and recalculation of feeders
Posted: Sun Dec 07, 2014 8:13 pm
by Steve Rowe
Nice, they kept that quiet, a genuine step forwards!
Re: Saving the rule file and recalculation of feeders
Posted: Mon Dec 08, 2014 9:46 am
by TrevorGoss
This does indeed sound useful, is there documentation on this new feature?
Thanks
Re: Saving the rule file and recalculation of feeders
Posted: Wed Feb 04, 2015 6:33 pm
by David Usherwood
We've dug into this a bit further - see
here. Key takeaway is that the old 'Spacebar save' no longer fires feeders - you need to use CubeProcessFeeders instead. On balance, an improvement I'd say.
Re: Saving the rule file and recalculation of feeders
Posted: Mon Feb 09, 2015 11:28 am
by TrevorGoss
We've dug into this a bit further - see here
Thanks for that Dave.
Key takeaway is that the old 'Spacebar save' no longer fires feeders - you need to use CubeProcessFeeders instead.
In performance Modeler, inside the properties tab of a TM1 service, there is an option called "Generate feeders automatically", this gives you a drop down list of 3 options: "No","Only automatically generated rules" and "All rules".
Are you aware of this and do you know if it makes a difference to the kicking of feeders. If I select "All rules", can choosing this option act as a replacement for using the CubeProcessFeeders function?
Thanks.
Trevor.
Re: Saving the rule file and recalculation of feeders
Posted: Mon Feb 09, 2015 1:27 pm
by mvaspal
Hi
I never used it but I would think the All Rules option generates feeder statements for your rule lines (including those manual rules that you wrote), and it is not for re-firing the existing feeders.
Re: Saving the rule file and recalculation of feeders
Posted: Mon Feb 09, 2015 4:01 pm
by tomok
TrevorGoss wrote:In performance Modeler, inside the properties tab of a TM1 service, there is an option called "Generate feeders automatically", this gives you a drop down list of 3 options: "No","Only automatically generated rules" and "All rules".
Are you aware of this and do you know if it makes a difference to the kicking of feeders. If I select "All rules", can choosing this option act as a replacement for using the CubeProcessFeeders function?
That's likely only going to be applicable to PM. How do I know? I don't for sure but common sense would tell me that for the TM1 service to know what to do during a rule save either the service would have a setting to tell it what to do or the client interface would. In this situation I know there is no setting in the TM1S.cfg file for this so the service isn't going to know automatically what to do so PM is going to be the one to tell it what to do. If you are doing all your development in PM then your OK. If you are doing a mixture then this isn't go to be a complete replacement for CubeProcessFeeders.
Re: Saving the rule file and recalculation of feeders
Posted: Mon Feb 09, 2015 4:39 pm
by TrevorGoss
Mvaspal and Tom,
thanks for both of your resposnes.
We do not develop in Performance Modeler, partly because of legacy, partly because it gets on our nerves.
If you are doing a mixture then this isn't go to be a complete replacement for CubeProcessFeeders.
You are most likely correct Tom, its not as if the CubeProcessFeeders function call is a real difficulty, especially if you run it as a start up chore, or a chore during the middle of the night.
Thanks.