Page 1 of 1
PA 2.0.6 Temporary Subset Unable To Register
Posted: Fri Dec 07, 2018 8:25 pm
by kenship
We just upgraded from 2.0.4 to 2.0.6 and realize many processes that utilize the automatic clearing of temporary view and subset not working for us.
It seems like if a process chain is called the clearing won't happen until the end of the process chain. In 2.0.4 it seems to be cleared at the end of each process.
We have a few process chains and within the chain the same dimension is re-engaged and now in 2.0.6 it couldn't proceed with it.
I'm now changing the TI to clear clear out the subset at the end of each process or to rename temporary subset different, but I wonder if someone else experience the same issue.
Thanks.
Kenneth
Re: PA 2.0.6 Temporary Subset Unable To Register
Posted: Sat Dec 08, 2018 11:57 pm
by Wim Gielis
Hi Kenneth,
I can relate to this problem. But I have no solution.
2 weeks ago after upgrading a TM1 model, I had to rewrite certain view and subset creations because of this issue with temporary objects. It was most of the time a mother process looping over 12 periods and within the loop a second process was called with a parameter for the period. After explicitly deleting the temporary views and subsets, the code ran just fine. It shouldn’t be necessary though.
Re: PA 2.0.6 Temporary Subset Unable To Register
Posted: Sun Dec 09, 2018 1:04 pm
by kenship
Thanks for your reply.
I just want to confirm that I'm not the only one who has this problem.
Ken
Re: PA 2.0.6 Temporary Subset Unable To Register
Posted: Sun Dec 09, 2018 1:21 pm
by Wim Gielis
kenship wrote: ↑Sun Dec 09, 2018 1:04 pmThanks for your reply.
I just want to confirm that I'm not the only one who has this problem.
Ken
Confirmed
And thanks for confirming my suspicions about the subject.
Re: PA 2.0.6 Temporary Subset Unable To Register
Posted: Sun Dec 09, 2018 3:34 pm
by lotsaram
I would suggest that a far better approach to deleting a temporary view or subset (which is a bit of an oxymoron) would be to ensure that all temp objects are assigned unique names.
e.g.
Code: Select all
cTempView = 'whatever';
cTimeStamp = TimSt( Now, '\Y\m\d\h\i\s' );
sRandomInt = NumberToString( INT( RAND( ) * 1000 ));
sTempView = Expand('%cTempView%_%cTimeStamp%_%sRandomInt%');
Actually it is a longstanding gripe of mine that the implementation of temporary objects was done very poorly leading to a far reduced scope where they can be used and encouraging poor coding practices due to temp objects being accessible only to the immediate TI process which created them as opposed to persisting for the duration of a transaction (and therefore being available to library function processes which would make them much more useful!) There is likewise fairly long in the tooth
enhancement request for this which IBM declined. Seeing this change of behavior gives me some hope that IBM has reconsidered and might be on the path to making temporary views and subsets a worthwhile and useful feature. Well we can live in hope.
Re: PA 2.0.6 Temporary Subset Unable To Register
Posted: Sun Dec 09, 2018 8:56 pm
by kenship
Thanks for the reply.
I probably won't go as far as you but like I said I either apply view and subset clearing in epilog or name the subset uniquely depending on the situation.
Kenneth
lotsaram wrote: ↑Sun Dec 09, 2018 3:34 pm
I would suggest that a far better approach to deleting a temporary view or subset (which is a bit of an oxymoron) would be to ensure that all temp objects are assigned unique names.
e.g.
Code: Select all
cTempView = 'whatever';
cTimeStamp = TimSt( Now, '\Y\m\d\h\i\s' );
sRandomInt = NumberToString( INT( RAND( ) * 1000 ));
sTempView = Expand('%cTempView%_%cTimeStamp%_%sRandomInt%');
Actually it is a longstanding gripe of mine that the implementation of temporary objects was done very poorly leading to a far reduced scope where they can be used and encouraging poor coding practices due to temp objects being accessible only to the immediate TI process which created them as opposed to persisting for the duration of a transaction (and therefore being available to library function processes which would make them much more useful!) There is likewise fairly long in the tooth
enhancement request for this which IBM declined. Seeing this change of behavior gives me some hope that IBM has reconsidered and might be on the path to making temporary views and subsets a worthwhile and useful feature. Well we can live in hope.