StatefulModel does not handle falsy values

[ben hockey]

this simple case shows how StatefuleModel? changes null values to empty strings. the null should be retained.

var d = {prop: null};
var m = new dojox.mvc.StatefulModel({data: d});
console.log(d.prop, m.toPlainObject().prop);

fyi - this is in the 1.7.0 release.

[ben hockey]

the same problem occurs for each of the properties with the following data:

var d = {
  u: undefined,
  z: 0,
  n: null,
  f: false
};
var m = new dojox.mvc.StatefulModel({data: d});
console.log(m.toPlainObject());

[ben hockey]

a possible fix is this (UPDATED):

StatefulModel.js

@@ -370 +370 @@
                        //              the data structure.
                        // tags:
                        //              private
-                       var data = (args && args.data) || this.data;
-                       if(data){
-                               this._createModel(data);
-                       }
+                       var data = (args && "data" in args) ? args.data : this.data;
+                       this._createModel(data);
                },
 
                //////////////////////// PRIVATE METHODS ////////////////////////
@@ -385 +383 @@
                        //              The input for the model, as a plain JavaScript object.
                        // tags:
                        //              private
-                       if(lang.isObject(obj) && !(obj instanceof Date) && !(obj instanceof RegExp)){
+                       if(lang.isObject(obj) && !(obj instanceof Date) && !(obj instanceof RegExp) && obj !== null){
                                for(var x in obj){
                                        var newProp = new StatefulModel({ data : obj[x] });
                                        this.set(x, newProp);

i would have added some tests but wasn't sure where to put them

[ben hockey]

just updated the diff in comment:3

[Ed Chatelain]

I will update a testcase for this and create a patch.

[Ed Chatelain]

Ben, Thanks for finding the problem and providing the fix. I just added the test into the date test where I had the test for the regexp. I also made an adjustment to the order that the robot tests were run, because this order seems to work more consistently for me.

[ben hockey]

ed, thanks for the patch.

i added one more test - for an empty string. since this case was not failing i had not mentioned it but i think it should be included in the tests so that the tests will reveal unexpected regressions.

once you can confirm that you're happy with the patch, i can commit it if you like.

[ben hockey]

hmm... just noticed a typo in my diff

doh.is("", test5_model.f.get('value'),"test5_model.s.get(value) should be an empty string");

should be

// was testing test5_model.f before!!!
doh.is("", test5_model.s.get('value'),"test5_model.s.get(value) should be an empty string");

the tests passed without any error so we might be getting false positives from doh so i'm going to look into changing the tests. i'll update my diff once i've changed the tests.

[ben hockey]

stricter assertions

[ben hockey]

ok, http://dojotoolkit.org/reference-guide/util/doh.html#doh-assertequal-obj1-obj2 says that doh.assertEqual (doh.is is an alias for doh.assertEqual) uses a fairly loose equality. i've updated the tests to use stricter equality.

[Ed Chatelain]

Hi Ben,

Your update looks good, it is also good to know about the loose equality for assertEqual, I will have to remember that. If you could commit it that would be great.

[ben hockey]

In [26940]:

add support for StatefulModel? to handle falsy values. includes tests.
fixes #14163 !strict

[cjolif]

In [27169]:

Fixes #14163 in 1.7.1 for edchat.