Normative: Add public and private class fields, static class features, private methods & getters/setters #1668

Ms2ger · 2019-08-14T13:49:07Z

This includes changes from https://tc39.es/proposal-static-class-features/.

ljharb · 2019-08-14T14:33:52Z

I’m confused; #1655 should contain all of this - “static class features” were merged back into the main proposal.

littledan · 2019-09-02T14:10:11Z

I've removed the "needs tests" label from this PR and the related ones as the tests are checked into test262.

ljharb · 2019-09-02T14:12:03Z

@littledan I’m looking for one PR to review for class fields; i see #1655, #1667, and this one.

littledan · 2019-09-02T14:29:28Z

Please review this one for the union of all three proposals.

spec.html

jmdyck

I also have a large pile of edits (at https://github.com/jmdyck/ecma262/tree/1668_ed) that I was planning to submit as a PR against this PR's branch, but github isn't letting me create a PR against the Ms2ger/ecma262 repo.

spec.html

ljharb · 2021-05-04T13:43:11Z

@danieltroger you can’t.

bakkot · 2021-05-07T03:03:54Z

Per discussion with @caiolima and @littledan at the editor call I've pushed commits addressing my previous three comments, as well as some other commits addressing most of the outstanding comments. As far as I'm aware, all that's left to be done is for @michaelficarra to suggest better wording than "unique", or decide he's fine with the current wording.

ljharb · 2021-05-07T03:52:54Z

spec.html

+        1. If _entry_.[[Kind]] is ~field~ or ~method~, then
+          1. Return _entry_.[[Value]].
+        1. Assert: _P_.[[Kind]] is ~accessor~.
+        1. If _P_.[[Get]] is *undefined*, throw a *TypeError* exception.


A normal accessor property returns undefined when no getter is present; is it intentional (and noticed in plenary/review) that a private accessor without a getter would throw?

That's been the case for the entire lifetime of the proposal, through a number of revisions and nearly four years of discussion. I'd certainly hope it was noticed by reviewers. I noticed, at any rate.

These semantics are what we got consensus on, and are now web reality, so if you'd like to propose different behavior a separate proposal is probably in order.

Fair enough. It could easily change since it's currently an error, but maybe there's no use case for it to do so.

IIRC, this is intentional based on the same design decision we made for throwing TypeError when a private field is not installed. I remember discussing in plenary about making this SyntaxError instead, since we already know at compile time that we are trying to apply [[Get]] for a not defined private getter, but we decided to keep current semantics. My memory might be tricking me, but I think the reason to throw TypeError is to make this proposal more friendly to a future decorators proposal.

spec.html

ljharb · 2021-05-07T03:56:51Z

spec.html

+      <h1>DefineMethodProperty ( _key_, _homeObject_, _closure_, _enumerable_ )</h1>
+      <p>The abstract operation DefineMethodProperty takes arguments _key_ (a property key or Private Name), _homeObject_ (an Object), _closure_ (a function object), and _enumerable_ (a Boolean). It performs the following steps when called:</p>
+      <emu-alg>
+        1. Perform SetFunctionName(_closure_, _key_).


Suggested change

1. Perform SetFunctionName(_closure_, _key_).

1. Perform ! SetFunctionName(_closure_, _key_).

there are a number of places where this AO is called without an appropriate prefix; i can put up a separate PR to fix those if editors are amenable.

I've added it here, but the prefix is not strictly necessary according to the spec's current rules (since it's just Performing, not consuming the value), so I'd prefer not to change the existing sites now and instead just fix them up as part of #1796.

It'd be necessary if it returned abruptly, no?

SetFunctionName cannot return abruptly.

sure, but i can't know that at the callsite (pending #1796)

spec.html

ljharb · 2021-05-07T04:32:42Z

spec.html

+        1. Let _newNames_ be a copy of _names_.
+        1. Append to _newNames_ the elements of PrivateBoundIdentifiers of |ClassBody|.


Suggested change

1. Let _newNames_ be a copy of _names_.

1. Append to _newNames_ the elements of PrivateBoundIdentifiers of |ClassBody|.

1. Let _newNames_ be a copy of _names_ with the elements of PrivateBoundIdentifiers of |ClassBody| appended.

this shorter form is used already in a few places in the spec.

Eh, I think the alias improves readability in this case.

spec.html

syg

lgtm, modulo question about what happens with [[ClassFieldInitializerName]] is a Private Name.

spec.html

syg · 2021-05-08T00:30:41Z

spec.html

+      <emu-note>
+        <p>For ease of specification, private methods and accessors are included alongside private fields in the [[PrivateElements]] slot of class instances. However, any given object has either all or none of the private methods and accessors defined by a given class. This feature has been designed so that implementations may choose to implement private methods and accessors using a strategy which does not require tracking each method or accessor individually.</p>
+        <p>For example, an implementation could directly associate instance private methods with their corresponding Private Name and track, for each object, which class constructors have run with that object as their `this` value. Looking up an instance private method on an object then consists of checking that the class constructor which defines the method has been used to initialize the object, then returning the method associated with the Private Name.</p>
+        <p>This differs from private fields: because field initializers can throw during class instantiation, an individual object may have some proper subset of the private fields of a given class, and so private fields must in general be tracked individually.</p>


Great note!

spec.html

jmdyck

Re:

PrivateElement vs PrivateElement Record
PrivateEnvironment vs PrivateEnvironment Record

For any given Record 'subtype', we generally (though with lots of exceptions) either consistently refer to it with the Record suffix, or consistently without.

spec.html

bakkot · 2021-05-08T22:02:57Z

PrivateElement vs PrivateElement Record

I've just pushed a commit which fixes these up to all be PrivateElement, to match its nearest cousin (Property Descriptor). Happy to go the other way if other editors prefer that though.

PrivateEnvironment vs PrivateEnvironment Record

As far as I can tell we're actually consistent about referring to these by "PrivateEnvironment Record". There is a pseudo -field named "PrivateEnvironment", but that isn't naming the type.

jmdyck · 2021-05-09T01:47:08Z

I've just pushed a commit which fixes these up to all be PrivateElement, to match its nearest cousin (Property Descriptor). Happy to go the other way if other editors prefer that though.

Looks good.

As far as I can tell we're actually consistent about referring to these by "PrivateEnvironment Record". There is a pseudo -field named "PrivateEnvironment", but that isn't naming the type.

Yup, my mistake.

caiolima · 2021-05-12T14:37:11Z

PR also LGTM. Thank you very much for all the work done there!

jmdyck

Minor changes suggested.

The last few commits fixed some other stuff I found.

Other than that, no complaints.

spec.html

…, private methods & getters/setters This includes changes from https://tc39.es/proposal-class-fields/. This includes changes from https://tc39.es/proposal-private-methods/. This includes changes from https://tc39.es/proposal-static-class-features/. Co-authored-by: Ms2ger <Ms2ger@gmail.com> Co-authored-by: Ujjwal Sharma <ryzokuken@igalia.com> Co-authored-by: Caio Lima <ticaiolima@gmail.com> Co-authored-by: Michael Dyck <jmdyck@ibiblio.org> Co-authored-by: Kevin Gibbons <bakkot@gmail.com> Co-authored-by: Jordan Harband <ljharb@gmail.com>

issues addressed

This comment has been minimized.

Sign in to view

Ms2ger force-pushed the static-fields branch from f79ae75 to 28147fe Compare August 15, 2019 10:13

This comment has been minimized.

Sign in to view

littledan removed the needs test262 tests The proposal should specify how to test an implementation. Ideally via github.com/tc39/test262 label Sep 2, 2019

This comment has been minimized.

Sign in to view

littledan added the has test262 tests label Sep 2, 2019

ljharb mentioned this pull request Sep 2, 2019

(ignore) Normative: Add public and private class fields #1655

Merged

ljharb changed the title ~~Normative: Add static fields.~~ Normative: Add public and private class fields Sep 2, 2019

ljharb requested review from zenparsing, waldemarhorwat, ljharb, littledan and a team September 2, 2019 14:45

ljharb reviewed Sep 2, 2019

View reviewed changes

spec.html Outdated Show resolved Hide resolved

spec.html Outdated Show resolved Hide resolved

ljharb requested a review from jmdyck September 2, 2019 14:55

ljharb changed the title ~~Normative: Add public and private class fields~~ Normative: Add public and private class fields, and static class features Sep 2, 2019

ljharb changed the title ~~Normative: Add public and private class fields, and static class features~~ Normative: Add public and private class fields, static class features, private methods & getters/setters Sep 2, 2019

ljharb mentioned this pull request Sep 2, 2019

(ignore) Normative: Add private methods and getter/setters. #1667

Merged

jmdyck requested changes Sep 5, 2019

View reviewed changes

ljharb reviewed May 7, 2021

View reviewed changes

jmdyck requested changes May 7, 2021

View reviewed changes

spec.html Outdated Show resolved Hide resolved

spec.html Outdated Show resolved Hide resolved

spec.html Show resolved Hide resolved

spec.html Show resolved Hide resolved

spec.html Show resolved Hide resolved

michaelficarra reviewed May 7, 2021

View reviewed changes

spec.html Show resolved Hide resolved

michaelficarra reviewed May 7, 2021

View reviewed changes

spec.html Show resolved Hide resolved

michaelficarra approved these changes May 7, 2021

View reviewed changes

syg reviewed May 8, 2021

View reviewed changes

jmdyck requested changes May 8, 2021

View reviewed changes

syg approved these changes May 10, 2021

View reviewed changes

ljharb requested a review from jmdyck May 12, 2021 15:04

jmdyck previously requested changes May 12, 2021

View reviewed changes

spec.html Outdated Show resolved Hide resolved

spec.html Show resolved Hide resolved

spec.html Show resolved Hide resolved

spec.html Show resolved Hide resolved

spec.html Show resolved Hide resolved

bakkot added the ready to merge Editors believe this PR needs no further reviews, and is ready to land. label May 12, 2021

ljharb force-pushed the static-fields branch from a5844ee to 56b5e8b Compare May 12, 2021 21:39

ljharb merged commit 56b5e8b into tc39:master May 12, 2021

This was referenced May 13, 2021

Editorial: Define+use list-concatenation #2382

Merged

Editorial: Rename some aliases #2353

Merged

aladdin-add mentioned this pull request May 27, 2021

Error trying to use top-level await? meriyah/meriyah#186

Closed

This was referenced Jul 17, 2021

Editorial: avoid duplicate SetFunctionName in {async,}generator methods #2463

Merged

Editorial: tweak PropertyDefinitionEvaluation and ClassElementEvaluation #2470

Merged

This was referenced Aug 29, 2021

Editorial: make the order of parameters for class field AOs more consistent #2504

Merged

Normative: add missing AsyncGenerator cases to ContainsArguments #2505

Merged

petamoriken mentioned this pull request Sep 27, 2021

Handle of private fields in the structured cloning whatwg/html#7123

Open

jmdyck mentioned this pull request Feb 11, 2022

Shouldn't ExecuteModule use the Execution Context constructed in InitializeEnvironment? #2660

Open

ljharb mentioned this pull request Apr 10, 2024

ECMAScript's OrdinaryFunctionCreate has another new parameter whatwg/html#10271

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Normative: Add public and private class fields, static class features, private methods & getters/setters #1668

Normative: Add public and private class fields, static class features, private methods & getters/setters #1668

Ms2ger commented Aug 14, 2019

This comment has been minimized.

ljharb commented Aug 14, 2019

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

littledan commented Sep 2, 2019

ljharb commented Sep 2, 2019 •

edited

Loading

littledan commented Sep 2, 2019

jmdyck left a comment

ljharb commented May 4, 2021

bakkot commented May 7, 2021

ljharb May 7, 2021

bakkot May 7, 2021 •

edited

Loading

ljharb May 7, 2021

caiolima May 12, 2021

ljharb May 7, 2021

bakkot May 7, 2021

ljharb May 7, 2021

bakkot May 7, 2021

ljharb May 7, 2021

ljharb May 7, 2021

bakkot May 7, 2021

syg left a comment

syg May 8, 2021

jmdyck left a comment

bakkot commented May 8, 2021

jmdyck commented May 9, 2021

caiolima commented May 12, 2021

jmdyck left a comment

	1. Perform SetFunctionName(_closure_, _key_).
	1. Perform ! SetFunctionName(_closure_, _key_).

		1. Let _newNames_ be a copy of _names_.
		1. Append to _newNames_ the elements of PrivateBoundIdentifiers of \|ClassBody\|.

	1. Let _newNames_ be a copy of _names_.
	1. Append to _newNames_ the elements of PrivateBoundIdentifiers of \|ClassBody\|.
	1. Let _newNames_ be a copy of _names_ with the elements of PrivateBoundIdentifiers of \|ClassBody\| appended.

Normative: Add public and private class fields, static class features, private methods & getters/setters #1668

Normative: Add public and private class fields, static class features, private methods & getters/setters #1668

Conversation

Ms2ger commented Aug 14, 2019

This comment has been minimized.

ljharb commented Aug 14, 2019

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

littledan commented Sep 2, 2019

ljharb commented Sep 2, 2019 • edited Loading

littledan commented Sep 2, 2019

jmdyck left a comment

Choose a reason for hiding this comment

ljharb commented May 4, 2021

bakkot commented May 7, 2021

Choose a reason for hiding this comment

bakkot May 7, 2021 • 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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

syg left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jmdyck left a comment

Choose a reason for hiding this comment

bakkot commented May 8, 2021

jmdyck commented May 9, 2021

caiolima commented May 12, 2021

jmdyck left a comment

Choose a reason for hiding this comment

ljharb commented Sep 2, 2019 •

edited

Loading

bakkot May 7, 2021 •

edited

Loading