345922 - Identifiers are reserved in attribute identifiers

Status: VERIFIED FIXED

(Core :: JavaScript Engine, defect, P3)

Description

[nanto_vi (TOYAMA Nao)]

From bug 336376 comment 6:
> > This is a proposed change for ECMA-262 Edition 4.  The suggestion comes from
> > Douglas Crockford.  Instead of reserving identifiers unconditionally, the lexer
> > would reserve outside of contexts that must be property names, specifically
> > 
> > * on the right of . and .. operators
> > * before the : in a property initialiser within an object initialiser
> 
>   * on the right of the :: operator where the right operand is an identifier

And should be on the right of @ operators.

For example,
  <label for="idref"/>.@for
should return "idref" instead of throwing syntax error.

Comment 1

[Blake Kaplan (:mrbkap) (inactive)]

Attachment: Fix

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Would there be any point in asserting that the bit is unset when you start?

Comment 3

[Mike Beltzner [:beltzner, not reading bugmail]]

Comment on attachment 230704 [details] [diff] [review]
Fix

a=drivers. Please land this on the MOZILLA_1_8_BRANCH.

We'd also like to know that we haven't the foggiest clue what you're doing here. :)

Comment 4

[Blake Kaplan (:mrbkap) (inactive)]

(In reply to comment #2)
> Would there be any point in asserting that the bit is unset when you start?

I looked into this, before attaching the patch, even. The bit is only set around calls to js_GetToken (and macros and functions wrapping it) and immediately unset, so it cannot possibly be set when we get here.

Comment 5

[Blake Kaplan (:mrbkap) (inactive)]

(In reply to comment #3)
> We'd also like to know that we haven't the foggiest clue what you're doing
> here. :)

One of the changes for JS1.7 is to allow keywords to be used as identifiers in certain places where keywords cannot possibly be used. For example, consider the fragment 'foo.if = 3'. The 'if' in the sample cannot possibly be an if statement, so we can simply treat it as the identifier "if" instead of the keyword. The original patch for that feature missed this particular case, which covers the E4X attribute selector (@), which this patch fixes.

Comment 6

[Blake Kaplan (:mrbkap) (inactive)]

Fix checked into the trunk and the 1.8 branch.

Comment 7

[Jeff Walden [:Waldo]]

Attachment: Update keyword unreserve tests

Comment 8

[Bob Clary [:bc] (inactive)]

Checking in regress-336376-01.js;
/cvsroot/mozilla/js/tests/js1_7/lexical/regress-336376-01.js,v  <--  regress-336376-01.js
new revision: 1.2; previous revision: 1.1

Comment 9

[Bob Clary [:bc] (inactive)]

verified fixed 1.8.1, 1.8 windows/mac(ppc|tel)/linux