Prerequisites
- [ X ] Put an X between the brackets on this line if you have done all of the following:
Description
#45, which supposedly fixed #11, introduced a regex that, in normal regex-land, should identify trailing commas after the last key-value pair in an object. This is strictly illegal syntax per the JSON spec: there is no trailing value-separator in 2.2 in https://www.ietf.org/rfc/rfc4627.txt , regardless of ignored whitespace. Node's implementation of the spec concurs:
> JSON.parse('{"a": 1,\n}')
SyntaxError: Unexpected token } in JSON at position 9
> JSON.parse('{"a": 1\n}')
{ a: 1 }
However, because Atom apparently feeds only one line at a time to the regex, and #45 handles newlines by placing them inside a lookahead group in a regex ... if there is a newline between the trailing comma and the end of the object, it will not be matched!
For anyone handcrafting JSON Schemas or config files (read: everyone these days) that will be read by strict parsers, this is a huge pain, especially since "trivial" config file changes that seem fine in an editor may break things in production, or only after a lengthy test run. More generally, we should strive to be compliant with the spec.
We may need to introduce a new scope in the grammar for the "space between a trailing comma and a closing }", rather than just having it as a lookahead. That way it would match over multiple lines.
Steps to Reproduce
In Atom console:
g = atom.grammars.grammarForScopeName('source.json'); JSON.stringify(g.tokenizeLines('{"a":1,\n}'), null, ' ')
Expected behavior: One token should have invalid.illegal.trailing-dictionary-separator.json scope.
Actual behavior: None of them do.
Reproduces how often: All the time.
Versions
atom --version
Atom : 1.17.0
Electron: 1.3.15
Chrome : 52.0.2743.82
Node : 6.5.0
apm --version
apm 1.17.0
npm 3.10.5
node 6.9.5 x64
python 2.7.12
git 2.7.1
OS 10.11.6
Additional Information
Any additional information, configuration or data that might be necessary to reproduce the issue.
Prerequisites
Description
#45, which supposedly fixed #11, introduced a regex that, in normal regex-land, should identify trailing commas after the last key-value pair in an object. This is strictly illegal syntax per the JSON spec: there is no trailing value-separator in 2.2 in https://www.ietf.org/rfc/rfc4627.txt , regardless of ignored whitespace. Node's implementation of the spec concurs:
However, because Atom apparently feeds only one line at a time to the regex, and #45 handles newlines by placing them inside a lookahead group in a regex ... if there is a newline between the trailing comma and the end of the object, it will not be matched!
For anyone handcrafting JSON Schemas or config files (read: everyone these days) that will be read by strict parsers, this is a huge pain, especially since "trivial" config file changes that seem fine in an editor may break things in production, or only after a lengthy test run. More generally, we should strive to be compliant with the spec.
We may need to introduce a new scope in the grammar for the "space between a trailing comma and a closing }", rather than just having it as a lookahead. That way it would match over multiple lines.
Steps to Reproduce
In Atom console:
g = atom.grammars.grammarForScopeName('source.json'); JSON.stringify(g.tokenizeLines('{"a":1,\n}'), null, ' ')Expected behavior: One token should have
invalid.illegal.trailing-dictionary-separator.jsonscope.Actual behavior: None of them do.
Reproduces how often: All the time.
Versions
atom --version
Atom : 1.17.0
Electron: 1.3.15
Chrome : 52.0.2743.82
Node : 6.5.0
apm --version
apm 1.17.0
npm 3.10.5
node 6.9.5 x64
python 2.7.12
git 2.7.1
OS 10.11.6
Additional Information
Any additional information, configuration or data that might be necessary to reproduce the issue.