Was surprised today that in Lucee, using cfloop query overwrites variables inside closures nested in the function. To explain: inside

loopArray

there is a closure on

arr

, which outputs

id

. It was expected to cascade the

id

value from

loopArray

, but the variable from cfloop was output. Coldfusion outputs a variable from

loopArray

. In Lucee, I repeated it on versions 5 and 6 at https://trycf.com/. Is it bug or I expect something wrong?

<cfscript>
qry = queryNew("id,name");
queryAddRow(qry);
querySetCell(qry,"id",1);
querySetCell(qry,"name","John");
queryAddRow(qry);
querySetCell(qry,"id",2);
querySetCell(qry,"name","Mary");

function loopArray(id) {
    var arr = [{
        "value": "100"
    }, {
        "value": "200"
    }];
    writeOutput("<br>id outside closure: " & id);
    arr.each(function(item){
        writeOutput("<br>id inside closure: " & id);
    });
}
cfloop (query=qry) {
    loopArray(id = 777);
};
</cfscript>

May be a bug but if you utilize this, it functions as I would expect it using the cfloop

I use

arr

only for make the closure example. The problem is that when I output "id" in function it's "777" (that value that I passed to function), but when I output "id" in nested closure it's "1" and "2" (value from cfloop). I expected value from parent function

But what "ID" are you expecting inside the loopArray and then within the closure?

The results will depend on how Lucee is handling scopes for the variables & the context in which

cfloop

,

loopArray

, and

arr.each

closure. Here, it seems as if Lucee is runing the

arr.each

closure function in the context `cfloop`/`qry` rather than in the context of

loopArray

. If you specify

arguments.id

in

loopArray

(instead of just

id

), will you get

777

for all of your responses? If not, then I'd consider it a bug. If it does, question of whether it is proper functionality is still open. One could argue it should maintain compatibility with ACF. (edit: use of

arguments.id

throws error that

id

doesn't exist in arguments scope, maybe because of context it's running in)

@Patrick Yes, I expect cascading context and

id

from parent (

loopArray

)

I think the closure side of things is at least partially a red herring here. The same thing can be demonstrated with far less complexity in play: https://trycf.com/gist/ddc0204c2aa4c533f54b47c8d7bdb58c/lucee6-beta?theme=monokai

<cfset qry = queryNew("id,name", "integer,varchar", [
    [1, "John"],
    [2, "Mary"]
])>
<cfset id = "outside query loop">
<cfoutput>Outside loop: #id#<br></cfoutput>


<cfloop query="qry">
    <cfset id = "inside query loop">
    <cfoutput>Inside loop: #id#<br></cfoutput><!--- 1,2 --->
</cfloop>

In a query loop... the first "scope" checked is the query reference. Whether or not it's in a closure should be neither here nor there. IMO Lucee's behaviour is correct here (the original). CF's is buggy.

To me, ACF's behavior seems more expected. Since the query variable ("qry") isn't actually declared inside the loopArray function, I wouldn't expect to see it pick any localized variables from inside the loop. For example, if you change the example to change the query column name from "id" to "id2" and then add an local variable called "id" either inside the cfloop() or outside, it prints 777 for every line: https://www.trycf.com/gist/302aaac9651d5ce7286e3fa84b84fe5e/lucee5 So it definitely looks like there's some weird cascading going on with the scoping that doesn't look right to me.

@Adam Cameron But in

loopArray

when I use

id

it calls from arguments scope (writeOutput("<br>id outside closure: " & id)), not from cfloop. I expect that closure will check its arguments scope, after that parents arguments scope and after that cloop scope. Why Lucee checks arguments for function, but doesn't check it for closure?

@dswitzer that's just because

id

(the arguments one) is not being shadowed by the same-named variable in the query any more. I did point that out. And this is entirely expected. I'm not sure what you are demonstrating with this. @romanstr good point! Despite that being clear in your code, I overlooked it due to being a div 😉. Yeah fair cop in that case. My "simplified" example is showing something different, missing that nuance.

Agree that Lucee should check arguments scope for closure but doesn't. Sounds like ACF does. And I imagine >80% of other languages would also. @romanstr Maybe worthwhile to post on dev.lucee.org and tag micstriit?

@Cage S I'll do it. Just wanted before to recheck myself and compare opinion with community 🙂

Don't tag Micha. If he wants to look at it, he will. I'd raise a ticket in Jira, and move on.

@Adam Cameron To me, the query variables should not be hoisted up to the closure inside a function. It doesn't do this with vars in the variables scope. That's what my example illustrates. I would not expect the closure inside the function to have any knowledge of query scoped variable if it cannot see vars in the variables scope in the same way.

Hehe, I'm sure he loves more work created for him 🙂 🙂

Being on the receiving end of random ppl tagging me for help on stuff, I am pretty bloody sure he doesn't.

Created bug report https://luceeserver.atlassian.net/jira/software/c/projects/LDEV/issues/LDEV-4771