Flattened the representation of compiled RegExp data.

All RegExp data are set on a single FixedArray instead of nesting them three deep.

[Lasse Reichstein, 2009-03-02 11:07:34]

Review, please.

[Erik Corry, 2009-03-02 12:01:48]

Does it lint and compile on ARM?

LGTM.

http://codereview.chromium.org/28311/diff/1001/1004
File src/jsregexp.cc (right):

http://codereview.chromium.org/28311/diff/1001/1004#newcode389
Line 389: // Ensures that the regexp object containst a compiled version of the
spolling!

http://codereview.chromium.org/28311/diff/1001/1004#newcode482
Line 482: int RegExpImpl::IrregexpNumberOfRegisters(Handle<FixedArray> re) {
This looks like exactly the same as IrregexpMaxRegisterCount

http://codereview.chromium.org/28311/diff/1001/1004#newcode759
Line 759: // TODO(lrn): How do we know the array is long enough.
Because we called EnsureSize at all the places where we could call here.  You
could ASSERT that it is true.

http://codereview.chromium.org/28311/diff/1001/1004#newcode774
Line 774: // Implmentation of the Irregexp regular expression engine.
Spolling!  (Not yours...)

[Lasse Reichstein, 2009-03-02 13:54:48]

Lint: yes.
Compile on ARM: Yes, now. I had actually broken the bytecode matching.

http://codereview.chromium.org/28311/diff/1001/1004
File src/jsregexp.cc (right):

http://codereview.chromium.org/28311/diff/1001/1004#newcode389
Line 389: // Ensures that the regexp object containst a compiled version of the
On 2009/03/02 12:01:48, Erik Corry wrote:
> spolling!

Done.

http://codereview.chromium.org/28311/diff/1001/1004#newcode482
Line 482: int RegExpImpl::IrregexpNumberOfRegisters(Handle<FixedArray> re) {
It is. Or it was, not it's gone.

http://codereview.chromium.org/28311/diff/1001/1004#newcode759
Line 759: // TODO(lrn): How do we know the array is long enough.
Done.