Notifier probe returning different data on alternate executions

[Neil Pate]

Hi all,

 

I have seen something today I was not expecting.

 

If I probe a notifier wire, in the probe window I get two different messages in the Probe Display on alternate executions of the same piece of code:

 

I get this:

Notifier name: 

Latest notification: 

Number waiting: 0

Refnum (in hex): 0x08800000

 

and then I get an empty value in the probe display, with the value in the table reading: "Unnamed - 0 waiting - 0x880000

 

So both tell me the same thing, but I wonder why I get this alternating effect. It is very noticeable in the probe window as the blue line that highlights the current probe is skinny and then fat.

 

This is LV2015.

 

Am I doing something strange in my code or have others seen a similar effect.

[ned]

I'd check your code very carefully. I've seen something like this before and finally traced it to a difficult-to-find coding error. I don't remember the exact details, but my vague recollection is I had some code that obtained a new notifier if the existing notifier input was invalid, coupled with a situation where I wasn't properly passing the notifier ref out only when the the existing notifier was valid (could have been a "use default if unwired" sort of thing). As a result, a probe on the wire would rapidly alternate between a valid and invalid notifier refnum.

 

EDIT: and then I looked at your images more closely... no idea. I don't have LV2015. Does every notifier refnum do that, or just one specific instance? You're not doing something funny like casting it to an int, then back to a notifier, right?

Edited November 11, 2015 by ned

[Neil Pate]

Yup I am pretty sure there is no other code interfering with this notifier. The refs are the same in both instances.

 

Does anybody know if we can poke inside the "standard" probes (i.e are they VIs like custom problems, or are they done at a lower level like inside the C++ code?) 

[OlivierL]

I have definitely seen the exact same behavior when probing a queue, in previous versions of LV. Sometime I would get a longer description and sometimes a shorter one, for the exact same wire. I'd be very interested to know if anyone from NI has an answer.