Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

suffix "opt" are missing in one line grammar (ex. 13.7.4.5) #908

Closed
kotarondo opened this issue May 2, 2017 · 1 comment · Fixed by tc39/ecmarkup#165
Closed

suffix "opt" are missing in one line grammar (ex. 13.7.4.5) #908

kotarondo opened this issue May 2, 2017 · 1 comment · Fixed by tc39/ecmarkup#165

Comments

@kotarondo
Copy link

Summary

For example, in 13.7.4.5
current draft and ES7 displays one line grammar as

IterationStatement : for ( LexicalDeclaration Expression ; Expression ) Statement
    Return the VarDeclaredNames of Statement.

If it is read literally, it does not define anything on the following grammar.

IterationStatement : for ( LexicalDeclaration ; ) Statement
IterationStatement : for ( LexicalDeclaration Expression ; ) Statement
IterationStatement : for ( LexicalDeclaration ; Expression ) Statement

This is not what we intend.

On the other hand, ES6 displays:

IterationStatement : for ( LexicalDeclaration Expression opt ; Expression opt ) Statement
    Return the VarDeclaredNames of Statement.

This is more comprehensive because "opt" is expanded as shown in "5.1.5 Grammar Notation".

NOTE1: How are "opt" suffixes dropped ?

Actually, in the raw html document, these Expressions are given opt suffix,
but they are not displayed because of "ecmarkup.css" line 249:

emu-production[collapsed] emu-mods {
  display: none;
}

Here "collapsed" attributes on "emu-production" are given in the build process.
Precisely, they are given in grammarkdown/out/lib/emitter/ecmarkup.js line 36:

    else if (node.body.kind === tokens_1.SyntaxKind.RightHandSide) {
        this.writer.write(" collapsed");
    }

where grammarkdown is a npm module called from ecmarkup build tool.

By the way, opt suffixes are DISPLAYED in multiline grammars (ex. 13.7.4.4)
because node.body.kind === tokens_1.SyntaxKind.RightHandSideList.

Anyway, I think the specification contents must be self-contained except for its displaying formats (fonts, etc). Outside format tools must not change the specification contents.

Note2

You can not say "opt suffix is automatically inserted when the corresponding Syntax has it."
because expanded grammars may be referred separately.

For example, In 13.3.2.4

VariableDeclaration : BindingIdentifier
    Return NormalCompletion(empty).

VariableDeclaration : BindingIdentifier Initializer
    Let bindingId be StringValue of BindingIdentifier.
    Let lhs be ? ResolveBinding(bindingId).
    ...

Here we can not insert opt suffix after Initializer
although the corresponding Syntax has it.

@kotarondo
Copy link
Author

Another example for Note2

In 13.3.3.2 Static Semantics: ContainsExpression

BindingElement: BindingPattern Initializer
    Return true.

Here you can insert 'opt' suffix after 'Initializer' without any formal contradiction on the spec, but it produces an unexpected specification.

jmdyck added a commit to jmdyck/ecmarkup that referenced this issue Dec 17, 2019
In practice, an `<emu-mods>` contains an `<emu-params>` and/or an `<emu-opt>`.
My guess is that this rule was introduced with only `<emu-params>` in mind,
as there are situations in which you might reasonably want to suppress
the display of `<emu-params>`, but you should never suppress an `<emu-opt>`.

---

An alternative to removing the rule
(and, if my above guess is correct, a solution
that would be more in line with the original intent)
would be to change its selector from:
  `emu-production[collapsed] emu-mods`
to:
  `emu-production[collapsed] emu-params`
I.e., in a collapsed production, suppress the display
of *only* the grammatical parameters (GPs).

However, I didn't propose that solution because it's incomplete in 3 ways:

- It doesn't suppress GPs in all contexts where we want to suppress them.
  (Generally we want to suppress GPs in any non-defining production,
  which includes all the collapsed production, but also includes lots of
  non-collapsed productions.)

- It doesn't suppress all the annotations we typically want to suppress
  (e.g. lookahead restrictions and no-LineTerminator-here annotations)
  because those don't occur in `<emu-mods>`.

- It doesn't allow for the rare cases where we *don't* want to suppress
  GPs (and annotations).

Instead (in ecma262 at least), we explicitly remove unnecessary GPs
and annotations from all productions where we don't want them to appear.
(Note that this also has the benefit of making the source easier to read.)
So even with the suggested modification, the rule would be basically pointless.

Better to just delete it.

---

This resolves issue tc39#108, and also:
- tc39/ecma262#536
- tc39/ecma262#908
- tc39/ecma262#1139
- tc39/ecma262#1663
ljharb pushed a commit to jmdyck/ecmarkup that referenced this issue Jan 17, 2020
In practice, an `<emu-mods>` contains an `<emu-params>` and/or an `<emu-opt>`.
My guess is that this rule was introduced with only `<emu-params>` in mind,
as there are situations in which you might reasonably want to suppress
the display of `<emu-params>`, but you should never suppress an `<emu-opt>`.

---

An alternative to removing the rule
(and, if my above guess is correct, a solution
that would be more in line with the original intent)
would be to change its selector from:
  `emu-production[collapsed] emu-mods`
to:
  `emu-production[collapsed] emu-params`
I.e., in a collapsed production, suppress the display
of *only* the grammatical parameters (GPs).

However, I didn't propose that solution because it's incomplete in 3 ways:

- It doesn't suppress GPs in all contexts where we want to suppress them.
  (Generally we want to suppress GPs in any non-defining production,
  which includes all the collapsed production, but also includes lots of
  non-collapsed productions.)

- It doesn't suppress all the annotations we typically want to suppress
  (e.g. lookahead restrictions and no-LineTerminator-here annotations)
  because those don't occur in `<emu-mods>`.

- It doesn't allow for the rare cases where we *don't* want to suppress
  GPs (and annotations).

Instead (in ecma262 at least), we explicitly remove unnecessary GPs
and annotations from all productions where we don't want them to appear.
(Note that this also has the benefit of making the source easier to read.)
So even with the suggested modification, the rule would be basically pointless.

Better to just delete it.

---

This resolves issue tc39#108, and also:
- tc39/ecma262#536
- tc39/ecma262#908
- tc39/ecma262#1139
- tc39/ecma262#1663
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

1 participant