Normative: move proto out of annex b #2125

devsnek · 2020-08-05T00:10:45Z

No description provided.

spec.html

Jack-Works · 2020-08-05T02:23:10Z

Is it still optional for non-web platforms?

devsnek · 2020-08-05T02:34:28Z

@Jack-Works Object.prototype.__proto__ is normative optional due to security concerns but { __proto__: ... } is not.

Jack-Works · 2020-08-05T02:41:43Z

I don't find what benefit we can get from the __proto__ literal 🤔

devsnek · 2020-08-05T02:42:38Z

@Jack-Works we have previous consensus to move items out of annex b and into the main spec where possible.

ljharb · 2020-08-05T04:31:59Z

spec.html

+        </emu-clause>
+      </emu-clause>
+
+      <emu-clause id="sec-object.prototype.__defineGetter__">


Why would we want the legacy define methods to be normative required??

The vast majority of annex b should remain normative optional, and anything that’s definitely only needed on the web imo should stay in annex b.

🤷 no one had any reason to make these ones optional, unlike the __proto__ accessor. even quickjs and xs implement __defineGetter__ and whatnot.

The reason - one I hold for many parts of Annex B - is that these things are bad and should never have existed, and more importantly, are not universally used.

__proto__ is used everywhere, in non-web code as well. __defineGetter_ and friends are not.

__defineGetter__ and friends are sufficiently widely used that any implementation intending to run JavaScript as it exists in the world needs to implement them. Writing down those things is the whole point of having a standard. I don't see the point of omitting them.

Incidentally, that's also what I understood to be consensus when this was discussed in plenary.

@ljharb I know the history of __proto__. The point I was making is that there's parts of the JavaScript standard that we now, in hindsight, consider to be mistakes (even outside of Annex B). Whether they were intentionally designed or organically became part of the standard after first becoming a de facto standard doesn't matter. It feels weird to special-case features just because they used to be in Annex B.

FWIW, I wish __proto__ was completely removed from specs, and from any engine. To date it caused more troubles than it supposed to solve, and both Object and Reflect have dedicated methods that replace it already.

The only usage I kinda think is reasonable, is for one-off literals creation, and literally nothing else: {__proto__: Shared, counter: 0} ... that's it.

@ljharb so I just need to add a note that says we don't like them?

The __proto__ accessor and other %Object.prototype% __‑prefixed and suffixed properties should remain in Annex B, as they have the much saner {Object,Reflect}.{defineProperty,getOwnPropertyDescriptor,getPrototypeOf,setPrototypeOf} alternatives.

In fact, in some old engines the __lookup*__ functions will return an accessor even if it’s shadowed by a data property.

@mathiasbynens that's totally true, but the entire purpose of Annex B was always "this is a special case we don't want in the actual language, and we don't want people using, but is here for web compat". In other words, I'm saying that "being in Annex B" is already being special-cased, and i don't want that indiscriminately lost.

@devsnek just like "normative optional" is a special tag, i'd want some kind of "deprecated" tag, that referenced a big scary note saying these are bad and if you use them you should feel bad, yes.

spec.html

ExE-Boss · 2020-09-11T17:10:00Z

This PR currently has unresolved issues and merge conflicts that prevent merging it.

Also, the lint issues have been fixed in tc39/ecmarkup#239.

phoddie · 2020-09-11T17:53:21Z

...even quickjs and xs implement defineGetter and whatnot...

FWIW – we held out on supporting __defineGetter__ and __proto__ in XS for a very long time. If I recall correctly, it was test262 that drove us to implement them. We would be happy to have them remain in Annex B.

devsnek · 2020-09-11T19:02:23Z

in case it was not clear, they are still normative optional.

ExE-Boss · 2020-09-15T20:43:42Z

Since this has several merge conflicts, moves other legacy __‑prefixed properties of Object.prototype out of Annex B, instead of just the “__proto__ Property Names in Object Initializers” bit (and that’s currently done in a way that breaks JSON text, ObjectAssignmentPattern, CoverParenthesizedExpressionAndArrowParameterList and CoverCallExpressionAndAsyncArrowHead), I’ve rebased this and opened #2175.

devsnek · 2020-09-15T21:32:02Z

I will update this if there is reason to do so (after we discuss it at TC39). There is no reason to open another PR.

zeldajay

Direct removal is not the best solution, we should seek a better one.

ljharb · 2020-09-21T03:06:53Z

Summarizing today's consensus:

proto syntax: required, non-"icky"
proto accessor: optional, "icky"
lookup/define methods: optional, "icky"

where "icky" conveys the existing sentiment for everything in Annex B found here:

All of the language features and behaviours specified in this annex have one or more undesirable characteristics and in the absence of legacy usage would be removed from this specification.

Additionally, there was interest expressed in noting which optional things were encouraged to be implemented when an engine's goals are interop; versus which "icky" things are discouraged for practitioners to use.

(I'll update the PR's consensus label when it matches that consensus)

devsnek · 2020-09-21T03:13:26Z

Will update shortly...

ExE-Boss · 2020-09-21T05:03:17Z

@devsnek Feel free to use my #2175 normative/object-proto-property-intializer branch as a base.

phoddie · 2020-09-21T19:07:30Z

Commenting on the summary from @ljharb, ECMA-402 contains text which appears to define its use of "normative optional" to mean "Annex B":

This specification uses blocks demarcated as Normative Optional to denote the sense of Annex B in ECMA-262. That is, normative optional sections are required when the ECMAScript host is a web browser. The content of the section is normative but optional if the ECMAScript host is not a web browser.

With this PR, TC39 would be using "normative optional" for the first time. It would be preferable to have a consistent definition of "normative optional", if possible, between ECMA-262 and ECMA-402.

ExE-Boss · 2020-09-21T21:08:03Z

Alternatively, perform the move to use “normative optional” in a future PR and keep the legacy properties on %Object.prototype% in Annex B for now.

- co-opts prose from Annex B description - dfn legacy and add emu-not-ref where appropriate - use dash in IDs - don't <dfn> legacy; clauses are what is legacy Co-authored-by: Jordan Harband <ljharb@gmail.com> Co-authored-by: Michael Ficarra <mficarra@shapesecurity.com>

Co-authored-by: Gus Caplan <me@gus.host> Co-authored-by: Jordan Harband <ljharb@gmail.com> Co-authored-by: Michael Dyck <jmdyck@ibiblio.org>

jmdyck

Looks good. (I'll try not to think about this any more.)

spec.html

- co-opts prose from Annex B description - dfn legacy and add emu-not-ref where appropriate - use dash in IDs - don't <dfn> legacy; clauses are what is legacy Co-authored-by: Jordan Harband <ljharb@gmail.com> Co-authored-by: Michael Ficarra <mficarra@shapesecurity.com>

Co-authored-by: Gus Caplan <me@gus.host> Co-authored-by: Jordan Harband <ljharb@gmail.com> Co-authored-by: Michael Dyck <jmdyck@ibiblio.org>

devsnek · 2021-07-22T05:08:17Z

🎉

- co-opts prose from Annex B description - dfn legacy and add emu-not-ref where appropriate - use dash in IDs - don't <dfn> legacy; clauses are what is legacy Co-authored-by: Jordan Harband <ljharb@gmail.com> Co-authored-by: Michael Ficarra <mficarra@shapesecurity.com>

Co-authored-by: Gus Caplan <me@gus.host> Co-authored-by: Jordan Harband <ljharb@gmail.com> Co-authored-by: Michael Dyck <jmdyck@ibiblio.org>

devsnek force-pushed the __proto__-shenanigans branch 3 times, most recently from 9a7d36d to 3572cef Compare August 5, 2020 00:33

gibson042 reviewed Aug 5, 2020

View reviewed changes

spec.html Outdated Show resolved Hide resolved

spec.html Outdated Show resolved Hide resolved

ljharb reviewed Aug 5, 2020

View reviewed changes

ExE-Boss reviewed Aug 5, 2020

View reviewed changes

spec.html Outdated Show resolved Hide resolved

devsnek mentioned this pull request Aug 8, 2020

Using __proto__ in object literals nodejs/help#2901

Closed

bakkot mentioned this pull request Aug 18, 2020

Linting: misc improvements tc39/ecmarkup#239

Merged

This was referenced Sep 15, 2020

RFC: Support __proto__ literal in object initializers microsoft/TypeScript#38385

Open

Normative: Move __proto__ initializer out of Annex B #2175

Closed

zeldajay reviewed Sep 16, 2020

View reviewed changes

jridgewell mentioned this pull request Sep 21, 2020

September 2020 babel/proposals#69

Open

devsnek force-pushed the __proto__-shenanigans branch from 3572cef to c27a2e0 Compare September 24, 2020 17:30

ljharb force-pushed the __proto__-shenanigans branch from e5ebd15 to d952b3b Compare July 13, 2021 18:17

ljharb requested a review from jmdyck July 13, 2021 18:18

ljharb added a commit to devsnek/ecma262 that referenced this pull request Jul 13, 2021

Normative: move __proto__ out of annex b (tc39#2125)

ed383ac

Co-authored-by: Gus Caplan <me@gus.host> Co-authored-by: Jordan Harband <ljharb@gmail.com> Co-authored-by: Michael Dyck <jmdyck@ibiblio.org>

ljharb force-pushed the __proto__-shenanigans branch from d952b3b to ed383ac Compare July 13, 2021 18:18

jmdyck approved these changes Jul 13, 2021

View reviewed changes

syg approved these changes Jul 19, 2021

View reviewed changes

Jack-Works reviewed Jul 20, 2021

View reviewed changes

spec.html Outdated Show resolved Hide resolved

ljharb added the editor call to be discussed in the next editor call label Jul 21, 2021

michaelficarra approved these changes Jul 21, 2021

View reviewed changes

michaelficarra removed the editor call to be discussed in the next editor call label Jul 22, 2021

bakkot approved these changes Jul 22, 2021

View reviewed changes

bakkot added the ready to merge Editors believe this PR needs no further reviews, and is ready to land. label Jul 22, 2021

ljharb and others added 2 commits July 21, 2021 21:55

Normative: move __proto__ out of annex b (tc39#2125)

1d70aa7

Co-authored-by: Gus Caplan <me@gus.host> Co-authored-by: Jordan Harband <ljharb@gmail.com> Co-authored-by: Michael Dyck <jmdyck@ibiblio.org>

ljharb force-pushed the __proto__-shenanigans branch from ed383ac to 1d70aa7 Compare July 22, 2021 04:55

ljharb merged commit 1d70aa7 into tc39:master Jul 22, 2021

devsnek deleted the __proto__-shenanigans branch July 22, 2021 05:08

jmdyck mentioned this pull request Jul 29, 2021

Editorial: Change |PropertyDefinition| to |ObjectLiteral| #2473

Merged

andreubotella mentioned this pull request Oct 11, 2021

basic node compatibity mode denoland/deno#12295

Closed

7 tasks

Jack-Works mentioned this pull request Apr 23, 2022

Support __proto__ in object literal (type-check only) microsoft/TypeScript#48816

Closed

y1d7ng mentioned this pull request May 3, 2022

assert: fix CallTracker calls wraps the function causes the original … nodejs/node#42909

Merged

Josh-Cena mentioned this pull request May 15, 2022

Content suggestion: Mention __proto__ in object literal initializers in the "Inheritance and the prototype chain" tutorial mdn/content#16068

Closed

bakkot mentioned this pull request May 17, 2022

Using __proto__ in object literals is not deprecated javascript-tutorial/en.javascript.info#3014

Closed

ghost mentioned this pull request Jul 19, 2022

fix(ext/dom_exception): align better with spec denoland/deno#15097

Merged

hyp3rflow mentioned this pull request Aug 3, 2022

fix: Update Object.prototype.__proto__ related comments denoland/deno#15394

Merged

petamoriken mentioned this pull request Mar 25, 2023

fix(webidl): do not throw if Object.prototype has been mutated denoland/deno#16407

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Normative: move proto out of annex b #2125

Normative: move proto out of annex b #2125

devsnek commented Aug 5, 2020

Jack-Works commented Aug 5, 2020

devsnek commented Aug 5, 2020

Jack-Works commented Aug 5, 2020

devsnek commented Aug 5, 2020

ljharb Aug 5, 2020

devsnek Aug 5, 2020 •

edited

Loading

mathiasbynens Aug 5, 2020

ljharb Aug 5, 2020

bakkot Aug 5, 2020

mathiasbynens Aug 5, 2020

WebReflection Aug 5, 2020

devsnek Aug 5, 2020

ExE-Boss Aug 5, 2020 •

edited

Loading

ljharb Aug 5, 2020

ExE-Boss commented Sep 11, 2020

phoddie commented Sep 11, 2020

devsnek commented Sep 11, 2020

ExE-Boss commented Sep 15, 2020 •

edited

Loading

devsnek commented Sep 15, 2020

zeldajay left a comment •

edited

Loading

ljharb commented Sep 21, 2020 •

edited

Loading

devsnek commented Sep 21, 2020

ExE-Boss commented Sep 21, 2020 •

edited

Loading

phoddie commented Sep 21, 2020

ExE-Boss commented Sep 21, 2020

jmdyck left a comment

devsnek commented Jul 22, 2021

Normative: move __proto__ out of annex b #2125

Normative: move __proto__ out of annex b #2125

Conversation

devsnek commented Aug 5, 2020

Jack-Works commented Aug 5, 2020

devsnek commented Aug 5, 2020

Jack-Works commented Aug 5, 2020

devsnek commented Aug 5, 2020

Choose a reason for hiding this comment

devsnek Aug 5, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ExE-Boss Aug 5, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ExE-Boss commented Sep 11, 2020

phoddie commented Sep 11, 2020

devsnek commented Sep 11, 2020

ExE-Boss commented Sep 15, 2020 • edited Loading

devsnek commented Sep 15, 2020

zeldajay left a comment • edited Loading

Choose a reason for hiding this comment

ljharb commented Sep 21, 2020 • edited Loading

devsnek commented Sep 21, 2020

ExE-Boss commented Sep 21, 2020 • edited Loading

phoddie commented Sep 21, 2020

ExE-Boss commented Sep 21, 2020

jmdyck left a comment

Choose a reason for hiding this comment

devsnek commented Jul 22, 2021

Normative: move proto out of annex b #2125

Normative: move proto out of annex b #2125

devsnek Aug 5, 2020 •

edited

Loading

ExE-Boss Aug 5, 2020 •

edited

Loading

ExE-Boss commented Sep 15, 2020 •

edited

Loading

zeldajay left a comment •

edited

Loading

ljharb commented Sep 21, 2020 •

edited

Loading

ExE-Boss commented Sep 21, 2020 •

edited

Loading