Nested Case statements

[Ugo DI LUCA]

Hi,

I was wondering what could be the difference from these 2 Case Statement structures. Is one more efficient that the other ? Why ?

Case(Value = "A", "Result 1", Value = "B", "Result 2", Value = "C", "Result 3")

Case(Value = "A", "Result 1", Case(Value = "B", "Result 2", "Result 3"))

Just curious about it

[danjacoby]

The difference is if the Value="D".

In the first example, you get nothing; in the second, you get "Result 3".

Flippant of me, I know.

[Ugo DI LUCA]

Thanks Dan, I knew I was missing something obvious.

I thought the empty result would have come with :

Case(Value = "A", "Result 1", Value = "B", "Result 2", Value = "C", "Result 3", " "), that is the last " "....

Thanx again.

[cjaeger]

as for timing - the nested case seems to be a little bit slower ...

don't know why.

just see attached test file (& correct the calc, that did escape me, dan ..)

caseTest.zip

[cjaeger]

had to delete sample records because of upload limit - have a look at the script to create new ones ...

[Ugo DI LUCA]

Hi Christian,

Thank you to reply in depth to my question.

Yes, apart from my errors about the empties, I had tested both methods while creating the calcs for the "Minimum Quantity in Stock" (you surely remember this thread), and I also had the feeling that a nested could be slower from a classic case.

Now, I don't have any idea of what I should see on your sample. My apologizes ;

[cjaeger]

>Now, I don't have any idea of what I should see on your sample. My apologizes

Instructions:

Open case-test.fp5

run script "make records" - this will create 20,000 records

the run script "run tests-start"

or just have a look at the results - a negative av(erage) means that case statement # 1 is faster than nested case statement #2.

The file replaces the field "test" using both case statements, recording script start time & end time before each run. In case results the time needed to run the replacement are recorded...

[paulage77]

It seems the nested case would be slower as the resulting abstract syntax tree would be wider/deeper to traverse.

[Ugo DI LUCA]

Well, in my mind, the nested would have eliminated first cases and concentrates the calcs only on the residual "true" results. This seems to me a quicker process.

Not sure I make me clear, but I still think it's strange.

BTW Christian....great test !

[paulage77]

That is a good point. What I mean, is that the resulting data structure for any nested conditional is larger than for single conditionals. More branches to look at and evaluate.