Determine whether a variable exists vs. is empty

[BAleiHi]

I need to determine whether a variable exists as distinct from whether it is empty.  From what i can tell, if i access a non-existent variable, it returns empty.  If $$foo doesn't exist, IsEmpty($$foo) returns true.  But If $$foo exists and is empty, then IsEmpty($$foo) also returns true.  How can I distinguish an existing but empty $$foo from a non-existant $$foo?  Thanks,

[comment]

An empty variable does not exist.

[bruceR]

Quote

"An empty variable does not exist."

Sort of.

Declare $$foo, with a value.

Use variable watcher (or script) to set $$foo to empty

$$foo is still visible in the variable watcher.

You can even close the variable watcher and open it again, and still see $$foo and its (now empty) value.

But I agree 100% with your point. Treat it as not-existing if empty.

[comment]

1 hour ago, BruceR said:

Use variable watcher (or script) to set $$foo to empty

$$foo is still visible in the variable watcher.

That's not what I observe. I have Script A to set a $$variable to a value, and Script B to set it to empty. Watching the Current tab of the Data Viewer, the $$variable comes up when Script A runs and is taken off the board when Script B runs.

[Mike Duncan]

I would guess there is in implicit "unset" type function that destroys the variable when it gets cleared via a script, but that behavior is not included when manually setting variables using the data viewer. The expected behavior is that setting a variable to "" will destroy it.

[Fitch]

It's also kind of interesting to create a couple of caclulations in the data viewer Watch panel:

Let( $$TESTMODE = "" ; "" )

Let( $$TESTMODE = 0 ; "" )

Whether or not the variable appears in the Current panel will depend on the order of the calcs in the Watch panel.

Be that as it may, I was able to replicate Bruce's result by manually clearing the variable in the data viewer.

[comment]

11 minutes ago, Mike Duncan said:

that behavior is not included when manually setting variables using the data viewer

 

10 minutes ago, Fitch said:

I was able to replicate Bruce's result by manually clearing the variable in the data viewer

This could be a feature of the data viewer. If you manually set the variable to empty, you may also wish to set it to something else later - so the data viewer keeps a spot for it. I don't think it has implications beyond that. It's sort of like "if a tree falls in a forest, and only the data viewer is around to hear it".

Edited November 6, 2015 by comment

[LaRetta]

Data Viewer can throw off perceptions in this way (and a few others).  Just running a script, the empty global variable disappears but if you've copied it to watch tab, it will stick around (only in the watch tab) even after script end.

I like that analogy, Michael.  :-)

 

And I mean setting it via script to empty - it still sticks around.  It doesn't have to be manually cleared.

[bruceR]

A little more detail. 

If I set up this expression in data viewer, evaluate it, then cancel, 

(so my expression is not stored in WATCH)

Let( [
N = 1
; $$foo1 = 1
; $$foo2 = 2
; $$foo3 = 3
/*
; $$foo1 = ""
; $$foo2 = ""
; $$foo3 = ""
*/
];
N
)

I get the first of the attached results.

If I eliminate the comment markers and evaluate again, cancel ( so the expression is not stored in WATCH) then the empty values are declared, - see second screenshot.

 

 

[comment]

15 minutes ago, LaRetta said:

the empty global variable disappears but if you've copied it to watch tab, it will stick around (only in the watch tab) even after script end.

I don't think that applies here. The Watch tab keeps track of expressions, not of variables.

 

13 minutes ago, BruceR said:

If I eliminate the comment markers and evaluate again, cancel ( so the expression is not stored in WATCH) then the empty values are declared, - see second screenshot.

I am looking at your second screenshot and I don't see them.

Edited November 7, 2015 by comment

[bruceR]

Yes. That is the correct result. They do no exist.

[comment]

1 minute ago, BruceR said:

Yes. That is the correct result. They do no exist.

Okay then. I guess you meant to say they were erased, instead of declared.

[bruceR]

I guess so. We may want to come up with consistent terminology.

My attempt to summarize:

If a variable is declared to hold a value (by script, Dataviewer watch statement, or whatever); then you can see it in the Current tab of Data Viewer.

In Data Viewer, you can even clear the value; but the variable name will persist, and in data viewer you can go back and put a value in it.

That is, if nothing else ( WATCH expression, script, layout object expression re-declares it).

However, if a variable declaration statement ( WATCH expression, script step, layout-based variable declaration, etc) sets (declares? erases?) the value to empty, when you go back to Data Viewer CURRENT tab you won't see the variable name at all.

[comment]

I don't disagree, I just think that for all practical purposes it's sufficient to say that setting a variable to empty erases it .

Even if we were to concede the possibility of an empty variable "existing" somehow, there is no way I know of to distinguish it from a variable that doesn't exist (to answer the original question). So this is just a matter of "do you believe that empty variables exist?".

[bruceR]

Yes. Empty variables don't exist.

[Fitch]

Things that are not can't be.

(warning, NSFW)