Hello community,

I don't understand something. I did my test with v91, v95 and v99 and noticed the same behaviour :

In a cube view, if I put a percentage format for an element through  the cube format manager or through the format attribute of the dimension, I noticed a weird behaviour. It does also the same if I don't have a percentage format but if I use the shorcut "3,1%" directly in the cell.

For instance:

• If I write 5,2% and then i double click in the cell I will see 5,2000000000000002 (not ok!)
• If I write 4,1 it shows 4,0999999999999996 (not ok !)
• If I write 6,1 it shows 6,1 (ok !)

If I take out the format I can see the wrong values in PAW afterwards.

What the impact for my users ?

1 -  If the user double click in the cell they will see many digits and not the exact number they wrote. Thus, it's confusing.

2 - In my model, the user set up a rate on a measure called Cloture_FIP_Std and then run a TI process. The TI process will do some calculations and update the reporting cube. It will also archive this rate on a specific technical measure that will be hidden for the user but that helps me for conditional formatting purposes (Sent_Cloture_FIP_Std)

By using the conditional formating in the view, I let the user know if the rate shown is the one that had been used in the last calculations by comparing Cloture_FIP_Std with Sent_Cloture_FIP_Std

So .... if I write 5,2% it shows 5,2000000000000002 by double click in the cell and then the process will archive 5,2 and 5,2000000000000002 <> 5,2 so the formatting will let the user thinks that he has to run again the calculations  .. And because it's not complicated enough if i put the format #0.0000000000000000% it will show 5,2000000000000005 🤓

Is it a known issue ? Is there a rational explanation to this ? 

Thanks for your help,

Ludovic

Hello community,

I don't understand something. I did my test with v91, v95 and v99 and noticed the same behaviour :

In a cube view, if I put a percentage format for an element through  the cube format manager or through the format attribute of the dimension, I noticed a weird behaviour. It does also the same if I don't have a percentage format but if I use the shorcut "3,1%" directly in the cell.

For instance:

• If I write 5,2% and then i double click in the cell I will see 5,2000000000000002 (not ok!)
• If I write 4,1 it shows 4,0999999999999996 (not ok !)
• If I write 6,1 it shows 6,1 (ok !)

If I take out the format I can see the wrong values in PAW afterwards.

What the impact for my users ?

1 -  If the user double click in the cell they will see many digits and not the exact number they wrote. Thus, it's confusing.

2 - In my model, the user set up a rate on a measure called Cloture_FIP_Std and then run a TI process. The TI process will do some calculations and update the reporting cube. It will also archive this rate on a specific technical measure that will be hidden for the user but that helps me for conditional formatting purposes (Sent_Cloture_FIP_Std)

By using the conditional formating in the view, I let the user know if the rate shown is the one that had been used in the last calculations by comparing Cloture_FIP_Std with Sent_Cloture_FIP_Std

So .... if I write 5,2% it shows 5,2000000000000002 by double click in the cell and then the process will archive 5,2 and 5,2000000000000002 <> 5,2 so the formatting will let the user thinks that he has to run again the calculations  .. And because it's not complicated enough if i put the format #0.0000000000000000% it will show 5,2000000000000005 🤓

Is it a known issue ? Is there a rational explanation to this ? 

Thanks for your help,

Ludovic

Hi Ludovic,

This unfortunately is not something new and unlikely to go away either.

Have a look at this article explaining IEEE-754 standard for floating point doubles

Hello community,

I don't understand something. I did my test with v91, v95 and v99 and noticed the same behaviour :

In a cube view, if I put a percentage format for an element through  the cube format manager or through the format attribute of the dimension, I noticed a weird behaviour. It does also the same if I don't have a percentage format but if I use the shorcut "3,1%" directly in the cell.

For instance:

• If I write 5,2% and then i double click in the cell I will see 5,2000000000000002 (not ok!)
• If I write 4,1 it shows 4,0999999999999996 (not ok !)
• If I write 6,1 it shows 6,1 (ok !)

If I take out the format I can see the wrong values in PAW afterwards.

What the impact for my users ?

1 -  If the user double click in the cell they will see many digits and not the exact number they wrote. Thus, it's confusing.

2 - In my model, the user set up a rate on a measure called Cloture_FIP_Std and then run a TI process. The TI process will do some calculations and update the reporting cube. It will also archive this rate on a specific technical measure that will be hidden for the user but that helps me for conditional formatting purposes (Sent_Cloture_FIP_Std)

By using the conditional formating in the view, I let the user know if the rate shown is the one that had been used in the last calculations by comparing Cloture_FIP_Std with Sent_Cloture_FIP_Std

So .... if I write 5,2% it shows 5,2000000000000002 by double click in the cell and then the process will archive 5,2 and 5,2000000000000002 <> 5,2 so the formatting will let the user thinks that he has to run again the calculations  .. And because it's not complicated enough if i put the format #0.0000000000000000% it will show 5,2000000000000005 🤓

Is it a known issue ? Is there a rational explanation to this ? 

Thanks for your help,

Ludovic

Thanks George.  You are correct that this is a limitation with how floating point numbers are stored.  This same behaviour can be seen in other software, although I've noticed that Excel now does a good job of hiding this.  This is part of the reason why TM1 has the concept of raw and formatted values for each numeric cell.  When you double click into the cell in the PAW cube viewer you see the raw value that is stored by TM1.  The same can be seen in the cell context menu.  When you are looking at the face of the cube viewer you see the formatted value.  You can change the formatting in TM1 or the PAW cube viewer to hide the insignificant digits.

We could consider an enhancement to hide the raw value (e.g a a cube viewer property to only display the formatted value).  

Hi Ludovic,

This unfortunately is not something new and unlikely to go away either.

Have a look at this article explaining IEEE-754 standard for floating point doubles

Hello community,

I don't understand something. I did my test with v91, v95 and v99 and noticed the same behaviour :

In a cube view, if I put a percentage format for an element through  the cube format manager or through the format attribute of the dimension, I noticed a weird behaviour. It does also the same if I don't have a percentage format but if I use the shorcut "3,1%" directly in the cell.

For instance:

• If I write 5,2% and then i double click in the cell I will see 5,2000000000000002 (not ok!)
• If I write 4,1 it shows 4,0999999999999996 (not ok !)
• If I write 6,1 it shows 6,1 (ok !)

If I take out the format I can see the wrong values in PAW afterwards.

What the impact for my users ?

1 -  If the user double click in the cell they will see many digits and not the exact number they wrote. Thus, it's confusing.

2 - In my model, the user set up a rate on a measure called Cloture_FIP_Std and then run a TI process. The TI process will do some calculations and update the reporting cube. It will also archive this rate on a specific technical measure that will be hidden for the user but that helps me for conditional formatting purposes (Sent_Cloture_FIP_Std)

By using the conditional formating in the view, I let the user know if the rate shown is the one that had been used in the last calculations by comparing Cloture_FIP_Std with Sent_Cloture_FIP_Std

So .... if I write 5,2% it shows 5,2000000000000002 by double click in the cell and then the process will archive 5,2 and 5,2000000000000002 <> 5,2 so the formatting will let the user thinks that he has to run again the calculations  .. And because it's not complicated enough if i put the format #0.0000000000000000% it will show 5,2000000000000005 🤓

Is it a known issue ? Is there a rational explanation to this ? 

Thanks for your help,

Ludovic