TM1web: SUBNM with "living" subset

[Lukas Meyer]

Hello,
I have a rather unusual problem - another time tm1web does not behave as I would expect it to do.
Basically I have two SUBNM formulas, one of them (S1) is completely static - the user just chooses an element. Then he is supposed to press an Action Button (B1) - which will run a process AND go to the very same websheet (without remembering the title elements), closing the current one.
The process will update an user-specific subset (based on the element chosen in S1)
The other SUBNM (S2) is not static - it uses a named range for the user-specific subset (which is a static subset, but updated during the above mentioned process).
So after choosing from S1, clicking B1 and choosing from S2 (which works as intended), the user decides to start it all over - so choosing from S1 (works fine) clicking B1 (works fine) and attempting to choose from S2 -which fails since it still holds the values intended for the the previous selection of S1. Manually changing the subset from the (web) Subset Editor (to a random other and back) will result in "TM1APIDOTNET Exception: - Operation Failed"

My guess is, that switching to the very same workbook (and replacing the current one) does not work at all - since entering a value to a blank cell (no DB-formula there ) and pressing the button will not delete the value but display it ...

From my understanding replacing a workbook by itself does work, doesn't it?
Since my approach fails pathetically - can someone help me? Summarized: depending on a selection A a process P should run and create a subset B from which the user can choose an element.

Any help appreciated
Lukas

[lotsaram]

It is a bit difficult to say, I thought I knew what your problem was but the more I reread your post the more I am confused by what you are saying.

So rather than posting a hypothetical solution based on guesswork maybe some further questions will help you out ..

1. Is the pseudo-navigation replacing the sheet with itself necessary? By the sounds of it if all that is happening is a TI should update a subset which the sheet is referencing then a recalc/rebuild only should be enough, no navigation needed.
2. Does S2 need to be a SUBNM formula, if it is referencing a named range which is filled by the subset built/updated by the element selection in S1 then wouldn't a dropdown list with no formula suffice, of possibly a VLOOKUP, MATCH or INDEX formula?
3. Is the formula in S1 and S2 a "standard" SUBNM formula with hardcoded text arguments?

The reason I ask point 3 is that this is a well known issue (or divergence from Excel functionality) with SUBNM in TM1 web. If a SUBNM in Excel has references to the dimension name for example that are the result of a formula or fed from another range then in Excel it will not be "selectable" (that is double-click will not invoke subset editor, it will be read only). However in TM1 web the formula will still have the SUBNM arrows and clicking will allow element selection with subset editor - but after selection the formula will become hardcoded or static as the range or formula reference will have been converted to text. This might be your problem.

[Lukas Meyer]

Hello,

1. The pseudo-navigation is required to set some corresponding flags since the described process has to be repeated twice (so select, press, select, press, select - and the user must not forget the "press-steps"), furthermore the subnm-fields need to be reset - as you mentioned in 3. (although on second thought this isn't necessary anymore. But it didn't work in the first place either )
2. This is what I am doing now - reading the process-created subset and display the values by using the drop-down - but I don't like it
3. I know very well - that's why I pseudo-navigate

So I don't use the subsets anymore, instead I use dropdowns (I'll tell you later whether or not this worked )
From your experience - the pseudo-navigation should work and thus reset dynamically created SUBNM-formulas, right?

(Meanwhile I doubt that the "pseudo-navigation replacing the sheet with itself" works - but it SHOULD work. No fair! What's the point in "replace" if in fact it doesn't replace but update?)