some unexpected js behavior i wanted to note

Reflect.defineProperty and Object.defineProperty seem to differ in behavior slightly

In one case I got an error with Object.defineProperty but not Refelect.defineProperty

TypeError: Cannot redefine property: length

Another was that, as far as i know, the only way to transition a non-configurable but writable property descriptor to writable false, is via a Object.freeze

{ value: x, configurable: false, writable: true }

to

{ value: x, configurable: false, writable: false }

[erights]

Reflect.defineProperty and Object.defineProperty seem to differ in behavior slightly

The definition of Reflect.defineProperty is at https://tc39.es/ecma262/#sec-reflect.defineproperty as a trivial wrapper around the internal [[DefineOwnProperty]] method, which indicates success by returning true or false.

The definition of Object.defineOwnProperty is at https://tc39.es/ecma262/#sec-object.defineproperty as a trivial wrapper around the internal [[DefinePropertyOrThrow]] internal function at https://tc39.es/ecma262/#sec-definepropertyorthrow which is a trivial wrapper around the internal [[DefineOwnProperty]] method.

From these, the only difference should be how they indicate normal failure. Reflect.defineProperty indicates normal failure by returning false. Object.defineOwnProperty turns that false case into a thrown TypeError. They should have exactly the same success behavior.

(Note that there’s a PR around somewhere — that I am surprised hasn’t been merged yet — to allow Object.defineProperty, in one particular awful break-the-web compat case, to indicate failure by returning false rather than throwing a TypeError. This applies only to the browser global object, i.e., the so-called “WindowProxy object” (which is not a proxy). Attn @ljharb where is this and why hasn’t it been merged?)

At https://web.archive.org/web/20160603161655/http://wiki.ecmascript.org/doku.php?id=es3.1:attribute_states is my best explanation of the possible state transitions of an EcmaScript property. It explains the diagram:

In the case where Object.defineProperty gave you the TypeError and presumably changed nothing, did Reflect.defineProperty instead return false and also equally change nothing?

Object.freeze should only be equivalent to a call to Object.preventExtensions and multiple calls to Object.defineProperty. Both Object.defineProperty and Reflect.defineProperty should be just as effective as Object.freeze at causing a writable, non-configurable property to become non-writable, non-configurable.

If any of these don’t seem right, could you post some compact test cases?

[kriskowal]

Closing since I don’t see any remaining actionable work. Please reopen and clarify otherwise.