Error, the IF function does not handle the error correctly

[veb86]

If there is an error in one of the results of the IF formula condition, then the error is shown if the condition points to this result
Example
= IF(1>0;1;#N/A)  =>  result:1
= IF(0>1;1;#N/A)  =>  result:#N/A

[wp]

What did you expect? When the 1st argument is true, the formula returns the 2nd argument (1), otherwise the third argument (#N/A). That's what is happening in Excel and in FPSpreadsheet.

I don't understand how the posted sample project relates to this.

[veb86]

Sorry, I didn't explain it well.
Excel:
= IF(1>0;0;#N/A)  =>  result:0
= IF(0>1;0;#N/A)  =>  result:#N/A

FPSpreadsheet:
= IF(1>0;0;#N/A)  =>  result:#N/A
= IF(0>1;0;#N/A)  =>  result:#N/A

I made the example simpler and gave a screenshot

[wp]

Are you using a recent version of FPSpreadsheet (from ccr)? It now produces the same output as Excel. I spent lots of time during the last weeks to look at all these borderline cases and write unit tests for them.

[veb86]

I checked. The library has been updated to r9615.
I have attached a screenshot. My other errors have been fixed.

[wp]

My other errors have been fixed.

Does this mean that the IF bug is still present? It is not for me (using r9615, but this did not change anything related to this bug)

[veb86]

I think you have an uncommitted section of code. This happens, please check again.
My mentor also checked and he has the same error.

[wp]

Finally got it (recently, I had investigated error propagation in Excel, found that the first error found among the arguments usually determines the formula result, and had applied this "rule" also to the IF function). Fixed in r9620.

[veb86]

The error has been resolved. Thank you!