Check for empty repetitions.

Added check that bails out of a repetition when the body is empty.

[Erik Corry, 2009-01-08 11:10:16]

LGTM

http://codereview.chromium.org/17409/diff/1/4
File src/jsregexp.cc (right):

http://codereview.chromium.org/17409/diff/1/4#newcode3006
Line 3006: if (!has_minimum &&
If we have a minimum but we know that we didn't match zero length then we want
to emit the on_success node now with no further checks.  This seems to miss that
case.

http://codereview.chromium.org/17409/diff/1/4#newcode3021
Line 3021: // number first and skip the empty check if it's not enough.
Ideally if we matched empty we should 'zoom' up to the minimum, right?  I guess
that's just an optimization.

http://codereview.chromium.org/17409/diff/1/12
File src/regexp-macro-assembler.h (right):

http://codereview.chromium.org/17409/diff/1/12#newcode139
Line 139: // label if it is.
Or rather, if it is not?

[Christian Plesner Hansen, 2009-01-08 12:40:49]

http://codereview.chromium.org/17409/diff/1/4
File src/jsregexp.cc (right):

http://codereview.chromium.org/17409/diff/1/4#newcode3006
Line 3006: if (!has_minimum &&
Good point, fixed.

http://codereview.chromium.org/17409/diff/1/4#newcode3021
Line 3021: // number first and skip the empty check if it's not enough.
We should but I'm not quite sure how to do that.  We may have to push a
backtrack for each of the remaining repetition, if not in practice then at least
in effect.

http://codereview.chromium.org/17409/diff/1/12
File src/regexp-macro-assembler.h (right):

http://codereview.chromium.org/17409/diff/1/12#newcode139
Line 139: // label if it is.
Whoops, forgot to update the comment when changing the method.  Fixed.