Here's a new upload.  I cleaned up some stuff putting the details of the
quick check that has been performed into the GenerationVariant object
instead of having a pointer.  Since GenerationVariants are copied
shallowly with the expectation that the copies are independent
afterwards this seemed more healthy.  Also addressed review comments and
added more explanation in the code.


http://codereview.chromium.org/14194/diff/1/7
File src/interpreter-irregexp.cc (right):

http://codereview.chromium.org/14194/diff/1/7#newcode279
Line 279: ASSERT(sizeof(Char) == 1);
On 2008/12/18 09:52:26, Christian Plesner Hansen wrote:
> Could be a static check.

I think that would provoke the template implementation of the compilers
a little too much.

http://codereview.chromium.org/14194/diff/1/11
File src/jsregexp.cc (right):

http://codereview.chromium.org/14194/diff/1/11#newcode1949
Line 1949: v |= v >> 1;
On 2008/12/18 09:52:26, Christian Plesner Hansen wrote:
> I'm terribly annoyed that I can't think of a simpler way to do this.
In any
> case it needs a 'v |= v >> 16' to work.

It's only intended for use with 16 bit characters but it's easier to add
the extra or operation than to think of a name that indicates that.

http://codereview.chromium.org/14194/diff/1/11#newcode2586
Line 2586: *   \    F    |
On 2008/12/18 09:52:26, Christian Plesner Hansen wrote:
> Does it lint?

Funnily enough it does, yes.

http://codereview.chromium.org/14194/diff/1/5
File src/jsregexp.h (right):

http://codereview.chromium.org/14194/diff/1/5#newcode1020
Line 1020: virtual int EatsAtLeast(int recursion_depth);  // Returns 0.
On 2008/12/18 09:52:26, Christian Plesner Hansen wrote:
> Since loop choice nodes now know which alternative is their continue
node maybe
> we can get a less conservative estimate than this.

I think this has to wait until we know the minimum number of characters
a loop body can match.

http://codereview.chromium.org/14194

--~--~---------~--~----~------------~-------~--~----~
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
-~----------~----~----~----~------~----~------~--~---

Reply via email to