Editorial: Eliminate "present" and "absent" fields #2624

jmdyck · 2022-01-14T01:41:14Z

When processing the merge of PR #2620, I noticed that although it had taken care of _rec_.[[Field]] is present, there were also cases of _rec_.[[Field]] is absent that should probably be handled similarly. And then I noticed that that still left some references to "present" and "absent" fields. So this PR is to clean all that up. And the last commit is a bonus.

There's actually one place left that uses "present" when talking about a field, and that's in Table 9: PrivateElement Fields, where one of the column-headers is "Values of the [[Kind]] field for which it is present" (where "it" is the field that a given row is declaring). I couldn't think of a good way to change that to the "record has a field" wording.

ljharb · 2022-01-14T02:39:51Z

spec.html

@@ -12640,13 +12643,13 @@ <h1>
            1. If ! IsDataDescriptor(_current_) is *true* and ! IsAccessorDescriptor(_Desc_) is *true*, then
              1. If _Desc_ has a [[Configurable]] field, let _configurable_ be _Desc_.[[Configurable]]; else let _configurable_ be _current_.[[Configurable]].
              1. If _Desc_ has a [[Enumerable]] field, let _enumerable_ be _Desc_.[[Enumerable]]; else let _enumerable_ be _current_.[[Enumerable]].
-              1. Replace the property named _P_ of object _O_ with an accessor property having [[Configurable]] and [[Enumerable]] attributes set to _configurable_ and _enumerable_, respectively, and each other attribute set to its corresponding value in _Desc_ if present, otherwise to its <emu-xref href="#table-object-property-attributes">default value</emu-xref>.
+              1. Replace the property named _P_ of object _O_ with an accessor property whose [[Configurable]] and [[Enumerable]] attributes are set to _configurable_ and _enumerable_, respectively, and whose [[Get]] and [[Set]] attributes are set by _Desc_ or default.


I’m not sure this is clear enough. Why not s/is present/if it has the field?

I’m not sure this is clear enough.

Do you mean:

set by _Desc_ or default doesn't clearly suggest its meaning (in which case we can look for a different phrase), or

it doesn't strongly signal that it's a use of a defined phrase (in which case we could consider a more distinctive syntax), or

the definition itself isn't clear enough, or

something else?

Why not s/if present/if it has the field/?

You could, but:

These steps are already fairly complicated syntactically; adding a subordinate clause will make them a bit more so.

Although it should be fairly clear that "it" refers to _Desc_, it's between two "its" that refer to the attribute, so that might be confusing.

Strictly speaking, "the field" wouldn't have an antecedent, although you could give it one by changing "the corresponding value" to "the value of the corresponding field".

I thought that factoring out the concept was useful.

All of those - "or default" isn't clear enough, doesn't have a definition, and thus doesn't clearly suggest its meaning.

Maybe it would be clearer to make a small AO that takes two records and the field name, and performs the logic? This prose has always been a bit unwieldy.

All of those - "or default" [...] doesn't have a definition,

You didn't see the definition in the AO's description?

Maybe it would be clearer to make a small AO that takes two records and the field name, and performs the logic? This prose has always been a bit unwieldy.

That would involve an alias/parameter whose value is a field name, and using it to:

check whether a given record (_Desc_) has a field of that name, and

if so, extract its value.

...which are things that the spec currently doesn't do. We could add such a thing, but an early version of PR #2468 tried something similar ("dynamic record field lookup", "record field name indirection"), and it met with some reluctance/opposition.

The spec passes around internal slot names as reified values, and has field names in tables as reified values.

(but totally fair about the dynamic lookup part)

michaelficarra · 2022-01-19T23:10:42Z

spec.html

@@ -12612,20 +12615,20 @@ <h1>
        </h1>
        <dl class="header">
          <dt>description</dt>
-          <dd>It returns a Boolean value which is *true* if and only if _Desc_ can be applied as the property of an object with specified _extensibility_ and current property _current_ while upholding <emu-xref href="#sec-invariants-of-the-essential-internal-methods">invariants</emu-xref>. When such application is possible and _O_ is not *undefined*, it is performed for the property named _P_ (which is created if necessary).</dd>
+          <dd>It returns a Boolean value which is *true* if and only if _Desc_ can be applied as the property of an object with specified _extensibility_ and current property _current_ while upholding <emu-xref href="#sec-invariants-of-the-essential-internal-methods">invariants</emu-xref>. When such application is possible and _O_ is not *undefined*, it is performed for the property named _P_ (which is created if necessary). For the purposes of this abstract operation, when we say an attribute is "set by _Desc_ or default", it means that if _Desc_ has a field of the same name as the attribute, the attribute is set to the value of that field; otherwise the attribute is set to its <emu-xref href="#table-object-property-attributes">default value</emu-xref>.</dd>


In the editor call today, we discussed this phrasing, and we do not think it adds clarity. Can you revert this and its usages? We think the original phrasing was clear enough.

I eliminated the "set by Desc or default" phrase, replacing it with a consistent blend of the two phrasings in the status quo, plus @ljharb's "if it has the field" suggestion. See 0cd53f1 for the net effect.

michaelficarra · 2022-01-20T01:41:49Z

spec.html

@@ -12620,12 +12623,12 @@ <h1>
            1. If _extensible_ is *false*, return *false*.
            1. If _O_ is *undefined*, return *true*.
            1. If ! IsAccessorDescriptor(_Desc_) is *true*, then
-              1. Create an own accessor property named _P_ of object _O_ whose [[Get]], [[Set]], [[Enumerable]], and [[Configurable]] attribute values are described by _Desc_. If the value of an attribute field of _Desc_ is absent, the attribute of the newly created property is set to its <emu-xref href="#table-object-property-attributes">default value</emu-xref>.
+              1. Create an own accessor property named _P_ of object _O_ whose [[Get]], [[Set]], [[Enumerable]], and [[Configurable]] attributes are set to the value of the corresponding field in _Desc_ if it has the field, otherwise to the attribute's <emu-xref href="#table-object-property-attributes">default value</emu-xref>.


Suggested change

1. Create an own accessor property named _P_ of object _O_ whose [[Get]], [[Set]], [[Enumerable]], and [[Configurable]] attributes are set to the value of the corresponding field in _Desc_ if it has the field, otherwise to the attribute's <emu-xref href="#table-object-property-attributes">default value</emu-xref>.

1. Create an own accessor property named _P_ of object _O_ whose [[Get]], [[Set]], [[Enumerable]], and [[Configurable]] attributes are set to the value of the corresponding field in _Desc_ if it has the field, and to the attribute's <emu-xref href="#table-object-property-attributes">default value</emu-xref> otherwise.

Okay. I wonder if "or" would be better than "and" though.

"and" seems unambiguously incorrect; "or" seems like the right choice here

I'm fine with either.

I've changed "and" to "or" per @ljharb, and squashed the fixup commit.

Change: `_rec_.[[Field]] is absent` to: `_rec_ does not have a [[Field]] field` as a follow-up to PR tc39#2620.

…tor (tc39#2624) In the "Create" and "Replace" steps in ValidateAndApplyPropertyDescriptor, use consistent phrasing for getting a value from _Desc_ or the default. In the "Replace" steps, clarify the "each other attribute" wording by naming the other attributes explicitly. And instead of talking about a record field being "present" or "absent", use the "record has a field" phrasing that PR tc39#2620 settled on.

That is, eliminate the remaining references to a record's fields being "present" or "absent".

i.e., IsAccessorDescriptor, IsDataDescriptor, IsGenericDescriptor - Prefer "has a field" over "does not have a field". - Prefer simple conditions over conjunctions.

spec.html

Change: `_rec_.[[Field]] is absent` to: `_rec_ does not have a [[Field]] field` as a follow-up to PR tc39#2620.

…tor (tc39#2624) In the "Create" and "Replace" steps in ValidateAndApplyPropertyDescriptor, use consistent phrasing for getting a value from _Desc_ or the default. In the "Replace" steps, clarify the "each other attribute" wording by naming the other attributes explicitly. And instead of talking about a record field being "present" or "absent", use the "record has a field" phrasing that PR tc39#2620 settled on.

That is, eliminate the remaining references to a record's fields being "present" or "absent".

i.e., IsAccessorDescriptor, IsDataDescriptor, IsGenericDescriptor - Prefer "has a field" over "does not have a field". - Prefer simple conditions over conjunctions.

jmdyck · 2022-02-02T03:00:09Z

(force-pushed to squash the fixup-commit)

Change: `_rec_.[[Field]] is absent` to: `_rec_ does not have a [[Field]] field` as a follow-up to PR tc39#2620.

…tor (tc39#2624) In the "Create" and "Replace" steps in ValidateAndApplyPropertyDescriptor, use consistent phrasing for getting a value from _Desc_ or the default. In the "Replace" steps, clarify the "each other attribute" wording by naming the other attributes explicitly. And instead of talking about a record field being "present" or "absent", use the "record has a field" phrasing that PR tc39#2620 settled on.

That is, eliminate the remaining references to a record's fields being "present" or "absent".

i.e., IsAccessorDescriptor, IsDataDescriptor, IsGenericDescriptor - Prefer "has a field" over "does not have a field". - Prefer simple conditions over conjunctions.

jmdyck added the editorial change label Jan 14, 2022

ljharb reviewed Jan 14, 2022

View reviewed changes

michaelficarra added the editor call to be discussed in the next editor call label Jan 19, 2022

michaelficarra reviewed Jan 19, 2022

View reviewed changes

jmdyck force-pushed the is_absent branch from d2d9804 to 5847f10 Compare January 20, 2022 01:22

michaelficarra reviewed Jan 20, 2022

View reviewed changes

jmdyck force-pushed the is_absent branch from 5847f10 to 45bb26f Compare January 20, 2022 01:58

michaelficarra approved these changes Jan 20, 2022

View reviewed changes

jmdyck added a commit to jmdyck/ecma262 that referenced this pull request Jan 20, 2022

Editorial: Eliminate _rec_.[[Field]] is absent (tc39#2624)

dbd6b48

Change: `_rec_.[[Field]] is absent` to: `_rec_ does not have a [[Field]] field` as a follow-up to PR tc39#2620.

jmdyck added a commit to jmdyck/ecma262 that referenced this pull request Jan 20, 2022

Editorial: Eliminate "present" or "absent" fields (tc39#2624)

cc491a3

That is, eliminate the remaining references to a record's fields being "present" or "absent".

jmdyck force-pushed the is_absent branch from 45bb26f to f6a3db2 Compare January 20, 2022 19:08

ljharb requested review from syg, bakkot and a team January 20, 2022 19:11

bakkot reviewed Jan 20, 2022

View reviewed changes

spec.html Outdated Show resolved Hide resolved

bakkot added the establishes editorial conventions label Jan 31, 2022

bakkot approved these changes Jan 31, 2022

View reviewed changes

jmdyck added a commit to jmdyck/ecma262 that referenced this pull request Feb 2, 2022

Editorial: Eliminate _rec_.[[Field]] is absent (tc39#2624)

b75d9e3

Change: `_rec_.[[Field]] is absent` to: `_rec_ does not have a [[Field]] field` as a follow-up to PR tc39#2620.

jmdyck added a commit to jmdyck/ecma262 that referenced this pull request Feb 2, 2022

Editorial: Eliminate "present" or "absent" fields (tc39#2624)

08b09c6

That is, eliminate the remaining references to a record's fields being "present" or "absent".

jmdyck force-pushed the is_absent branch from 7e1d9c9 to f936b4e Compare February 2, 2022 02:58

syg approved these changes Feb 2, 2022

View reviewed changes

michaelficarra added ready to merge Editors believe this PR needs no further reviews, and is ready to land. and removed editor call to be discussed in the next editor call labels Feb 2, 2022

jmdyck added 2 commits February 16, 2022 15:18

Editorial: Eliminate _rec_.[[Field]] is absent (tc39#2624)

0b35749

Change: `_rec_.[[Field]] is absent` to: `_rec_ does not have a [[Field]] field` as a follow-up to PR tc39#2620.

jmdyck added 2 commits February 16, 2022 15:18

Editorial: Eliminate "present" or "absent" fields (tc39#2624)

b3c29fd

That is, eliminate the remaining references to a record's fields being "present" or "absent".

Editorial: Simplify IsFooDescriptor ops (tc39#2624)

1c7ae4b

i.e., IsAccessorDescriptor, IsDataDescriptor, IsGenericDescriptor - Prefer "has a field" over "does not have a field". - Prefer simple conditions over conjunctions.

ljharb force-pushed the is_absent branch from f936b4e to 1c7ae4b Compare February 16, 2022 23:18

ljharb merged commit 1c7ae4b into tc39:main Feb 16, 2022

jmdyck deleted the is_absent branch February 18, 2022 01:42

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Editorial: Eliminate "present" and "absent" fields #2624

Editorial: Eliminate "present" and "absent" fields #2624

jmdyck commented Jan 14, 2022

ljharb Jan 14, 2022

jmdyck Jan 14, 2022

ljharb Jan 14, 2022

jmdyck Jan 14, 2022 •

edited

Loading

ljharb Jan 14, 2022

ljharb Jan 14, 2022

michaelficarra Jan 19, 2022

jmdyck Jan 20, 2022 •

edited

Loading

michaelficarra Jan 20, 2022

jmdyck Jan 20, 2022

ljharb Jan 20, 2022

michaelficarra Jan 20, 2022

jmdyck Jan 20, 2022

jmdyck commented Feb 2, 2022

	1. Create an own accessor property named _P_ of object _O_ whose [[Get]], [[Set]], [[Enumerable]], and [[Configurable]] attributes are set to the value of the corresponding field in _Desc_ if it has the field, otherwise to the attribute's <emu-xref href="#table-object-property-attributes">default value</emu-xref>.
	1. Create an own accessor property named _P_ of object _O_ whose [[Get]], [[Set]], [[Enumerable]], and [[Configurable]] attributes are set to the value of the corresponding field in _Desc_ if it has the field, and to the attribute's <emu-xref href="#table-object-property-attributes">default value</emu-xref> otherwise.

Editorial: Eliminate "present" and "absent" fields #2624

Editorial: Eliminate "present" and "absent" fields #2624

Conversation

jmdyck commented Jan 14, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jmdyck Jan 14, 2022 • 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

jmdyck Jan 20, 2022 • 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

jmdyck commented Feb 2, 2022

jmdyck Jan 14, 2022 •

edited

Loading

jmdyck Jan 20, 2022 •

edited

Loading