VALUE? inconsistency

[rebolbot]

Submitted by: Ladislav

The VALUE? function served in R2 as a kind of guard against triggering an error when trying to get a value of a word.

This property made the function useful and it was also in agreement with the documentation stating that "Returns TRUE if the word has been set...Returns FALSE for all other values." (in R2) or "The VALUE? function returns TRUE if the specified word has a value. It returns FALSE if not." (in R3).

In both cases it is a consequence of logic that when the VALUE? function doesn't return TRUE it shall return FALSE.

; a word that "isn't bound to a context"
first to block! "x" ; == x
bound? first to block! "x" ; == none
; another word that "isn't bound to a context at present"
f: func [x] ['x]
f none ; == x
; now observe how the former is handled by VALUE?
value? first to block! "x" ; == false
; however, the latter is not handled in a compatible manner
value? f none
; ** Script error: x word is not bound to a context
; ** Where: value?
; ** Near: value? f none
; note that I do not disagree with the formulation;
; I disagree with the inconsistency

^{CC - Data [ Version: alpha 112 Type: Bug Platform: All Category: Native Reproduce: Always Fixed-in:none ]}

[rebolbot]

Submitted by: BrianH

This is a side effect of #1893. In the case of value? f none, the script error returned is incorrect. The word is not really not bound to a context, it's just bound to an out-of-scope function context. So, VALUE? isn't being inconsistent, the contexts are.

[rebolbot]

Submitted by: Ladislav

Yes, the "out-of-scope function context" text looks more accurate. However, my opinion is that the VALUE? function documentation specifies that the function shall return FALSE when it does not return TRUE. Also, I do not think it is a side-effect of #1893.

[rebolbot]

Submitted by: Ladislav

In the core-tests suite.

[rebolbot]

Submitted by: Ladislav

Pull request submitted.