John Hammond wrote:Code: Select all
------------------------------------- prolog tab --------------------
DimensionElementComponentAdd('dimChartOfAccounts', 'Motor Vehicles', 'John Hammond', 1.00);
------------------------------------- metadata tab --------------------
testit = DimensionElementComponentAdd('dimChartOfAccounts', v1, v1 | v2, 1.00);
xyz = STR(testit,6,0) ;
------------------------------------- data tab --------------------
asciioutput('c:\test.csv',v1 );
asciioutput('c:\test.csv',v2 );
asciioutput('c:\test.csv', '-----------------------------------' | xyz | '--------------------------------' );
This was very simple code to go thru a list of parent entries in the .csv input file and insert a child beneath them with a suffix.
1. When I used DimensionElementComponentAdd in the Data tab it did not work - but it did ni the Metadata tab.
That's correct, at least in 9.1 and presumably other versions as well.
However that's the metadata tab's purpose. You can get away with doing metadata actions on the Prolog or Epilog as well as well (or at least some of them) but at the end of the Metadata tab the system will do all of the necessary compilation and saving of the dimensions, attributes, etcetera - in short, it updates the database's metadata - which means that the cubes will then be ready to receive values in the Data tab. This doesn't happen in the Data tab itself. At least, that is what we've surmised by trial and error over the years; the documentation is its usual helpful self on such issues.
In other words, if it relates to writing data from the data source to the cube, it goes on the Data tab. If it involves updating cube or dimension structures, Metadata tab. Both tabs loop through the whole data source once (if there's any code in the relevant tab).
That's a better guide than trying to identify which individual functions should be used on which tab.
John Hammond wrote:
2. When I used asciioutput in the Metadata tab it did not work - but it did in the data tab.
Either there was no data to process on the time that you attempted that, or there may have been a bug in your code, because AsciiOutput
does work on any of the tabs. One thing to watch out for, though; if you have an AsciiOutput on (say) the Metadata tab, each row that you process from the data source will append a new line to that output file. That's not a problem.
However if you AsciiOutput to a text file
of the same name in a later tab (either the Data or Epilog tab) that output will blow away the file that you created on the Metadata tab. In other words, each tab will overwrite any output file that was created by an earlier tab. If you need to output from multiple tabs in your process, be sure to use different file names.
John Hammond wrote:
3. When I continue to try and add multiple elements as in the prolog - which I run again and again without error - I would expect there to be some kind of error mechanism to complain that I was trying to add duplicate elements.
However when I use testit to check if there is a return code then I always get I.
I know that DimensionElementComponentAdd is supposed to be a function, and functions are supposed to return values, but... don't rely on return values from any function
unless the documentation tells you that you can. Unless Iboglix explicitly states that a return value has a particular meaning, then (a) it may not have, and (b) there's no guarantee that they won't change what that meaning is in a future version without mentioning it.
It appears that if the element already exists in the consolidation then DimensionElementComponentAdd just ignores it. (Again tested in 9.1.) However this is consistent with the behaviour of most metadata functions, and is in fact quite useful. For example if you're updating a chart of accounts you don't need to be receiving error messages from the DimensionElementInsert function to tell you that elements already exist; if they do, they're just ignored, if they don't, they're added. It appears to be a similar sort of thing with the DimensionElementComponentAdd function. If it were otherwise the only way you could avoid the error would be by doing a DimIx test (ElIsComp in your case) to see whether the element already existed before you tried to add it. The way it has been implemented is faster, requires less coding on our part, and avoids unnecessary errors.
Good gods, did I just praise the way that Applix implemented something?!
Let me qualify that before the long term membership swoons; I grant you that they probably
could have coded the function so that it returned different codes depending on the result, but I don't think it's that critical; if you really need to check whether the element that you're adding is already there and write to an error log you can use ElIsComp in conjunction with either ItemReject or an AsciiOutput.
John Hammond wrote:
The problem that I might enter an invalid parent is correctly flagged in the message log, however.
Does anyone know whether TM1 has some sort of return code system
{Cough}, I'll leave that issue for your namesake Mr. Hobson...
John Hammond wrote:
and why and what functions are tab specific.
There aren't any that I can think of which really are tab specific as such, though they may not always work as advertised on all tabs. You can determine that from the context of what the function is supposed to do. As I said if you're processing from a data source you should keep metadata actions on the metadata tab. If you're not (data source of "None"), I'd generally only do them on the Prolog tab.
