>>>Gregory:
>>>
>>>as mentioned in the initial message:
>>>
>>>No behaviour change
>>>
>>>The record with this problem is deleted anyway and never comes up into the resultset
>>>
>>>BTW, it would be
>>>
>>>

>>>padr(nvl(field, SET("Nulldisplay")), 11)
>>>

>>>
>>>Agnes
>>>
>>>Update:
>>>
>>>Hm. I'm shure to have thisone tested, but now it gives right result with NVL. Need to re check
>>>
>>> Anyway, its wrong behaviour.
>>>
>>>?PADR(.NULL.,11) gives a result.
>>>
>>
>>PADR(NULL,n) returns NULL. So, you always need to convert NULL to a string.
>>
>>Need to check if this behavior (PADR(NULL) = NULL) is documented.
>
>Hm. something is strange or influenced by some SET command. I need to recreate the situation tomorrow.
>I'm shure that I have tested the SELECT with NVL (it's to obvious) and also get some meaningfull return from NVL(.null.,11)+'A'
>
>Anyway if the source is
>1
>.NULL.
>2
>
>it gives proper result, so shouldn't do
>
>.NULL.
>1
>2
>
>the same?
>
>IOW

I see the problem with NVL in my testing - I have adressed the wrong target in this test, so NVL works as expected by you and Gregory.

But it the it will return a string instead of .NULL. what is wrong again.