See bug 401466. That was a Narcissus bug, different from this one, but it exposed this SpiderMonkey bug. Given x=1, y=2: eval("(x\n)-- y") results in 2, instead of a SyntaxError being thrown (the eval is not essential, just makes it easier to show the newline). Without eval: js> x=1, y=2 2 js> (x )-- y 1 Note how the ; is inserted before the --. Without the newline: js> (x)-- y typein:4: SyntaxError: missing ; before statement: typein:4: (x)-- y typein:4: ......^ which follows the spec. /be

Results with the patch: js> x=1, y=2 2 js> (x )-- y typein:3: SyntaxError: missing ; before statement: typein:3: )-- y typein:3: ....^ /be

Shoulda thunk harder before posting that last patch. /be

Comment on attachment 287274 [details] [diff] [review] the right fix a+ schrep - is there a regression test for this?

(In reply to comment #3) > (From update of attachment 287274 [details] [diff] [review]) > a+ schrep - is there a regression test for this? > if in-testsuite+ is set, there is a test. There will be one when the bug is fixed.

Fixed: js/src/jsparse.c 3.307 /be

/cvsroot/mozilla/js/tests/ecma_3/LexicalConventions/7.9.1.js,v <-- 7.9.1.js initial revision: 1.1 /cvsroot/mozilla/js/tests/ecma_3/LexicalConventions/browser.js,v <-- browser.js initial revision: 1.1 /cvsroot/mozilla/js/tests/ecma_3/LexicalConventions/shell.js,v <-- shell.js initial revision: 1.1 /cvsroot/mozilla/js/tests/ecma_3/extensions/7.9.1.js,v <-- 7.9.1.js initial revision: 1.1 note the special treatment of throw in ecma_3/extensions/7.9.1.js due to bug 256617.

verified fixed 1.9.0 linux|mac|win