383574 - regexp /\0/ triggers bogus strict warning "non-octal digit in an escape sequence that doesn't match a back-reference"

Status: RESOLVED WORKSFORME

(Core :: JavaScript Engine, defect)

Description

[Jesse Ruderman]

js> options("strict")

js> "foo\0bar".replace(/\0/, "NULL")
typein:2: strict warning: non-octal digit in an escape sequence that doesn't match a back-reference:
typein:2: strict warning: "foo\0bar".replace(/\0/, "NULL")
typein:2: strict warning: ...................^
fooNULLbar

\0 seems to be a recommended way to search for null characters (see http://developer.mozilla.org/en/docs/Core_JavaScript_1.5_Reference:Global_Objects:RegExp), so I think this warning is bogus.

Comment 1

[Brian Crowder]

x00000000:  I think a patch of yours from another bug fixes this.  If I'm right, let's dup?

Comment 2

[x0]

Attachment: patch

(In reply to comment #1)
> x00000000:  I think a patch of yours from another bug fixes this.

No, bug 429243 looks similar, but isn't related.

The patch is not fully tested yet.

Regarding bug 306727 comment 12: Specified behavior for /\01/ would be the same as for /01/. \0 in a DecimalEscape cannot be followed by a DecimalDigit because a DecimalIntegerLiteral cannot start with 0, unless it is only the single digit 0, and treating it as /0x001/ is prohibited by the lookahead (although that would make most sense to me). That means it has to be parsed as IdentityEscape instead.

If Appendix B.1.2 would apply to RegExp literals (makes sense if it applies to strings), then /\08/ should still be /08/, and the octal digits eat never more than 3 digits, so /\0001/ should be /\x001/ like "\0001" is "\x001". /\001/ would be /\x01/, however.

Comment 3

[x0]

Attachment: patch that doesn't break /\100/

Comment 4

[Brian Crowder]

Comment on attachment 318087 [details] [diff] [review]
patch that doesn't break /\100/

Is this awaiting review, or does it need more testing?

Comment 5

[Ryan VanderMeulen (PTO, back July 27)]

js> "foo\0bar".replace(/\0/, "NULL")
"fooNULLbar"