Handlify DescriptorArray::Merge().

Handlify DescriptorArray::Merge().

R=svenpanne@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=20339

[Benedikt Meurer, 2014-02-17 08:22:37]

Hey Sven,
This is required for field type tracking. Pretty straight forward though.
PTAL
-- Benedikt

[Sven Panne, 2014-02-17 12:45:17]

LGTM with nits.

https://codereview.chromium.org/169363002/diff/50001/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/169363002/diff/50001/src/objects.cc#newcode8038
src/objects.cc:8038: Name* key = desc->GetKey(descriptor);
Can we substitute away this variable? I *think* it is safe to use a naked
pointer here, but it's far from obvious. Moving it closer to its use makes
things clearer and safer.

https://codereview.chromium.org/169363002/diff/50001/src/objects.cc#newcode8064
src/objects.cc:8064: Name* key = other->GetKey(descriptor);
Same here, even though we have 2 uses.

[Benedikt Meurer, 2014-02-17 13:15:22]

https://codereview.chromium.org/169363002/diff/50001/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/169363002/diff/50001/src/objects.cc#newcode8038
src/objects.cc:8038: Name* key = desc->GetKey(descriptor);
On 2014/02/17 12:45:17, Sven Panne wrote:
> Can we substitute away this variable? I *think* it is safe to use a naked
> pointer here, but it's far from obvious. Moving it closer to its use makes
> things clearer and safer.

Done.

https://codereview.chromium.org/169363002/diff/50001/src/objects.cc#newcode8064
src/objects.cc:8064: Name* key = other->GetKey(descriptor);
On 2014/02/17 12:45:17, Sven Panne wrote:
> Same here, even though we have 2 uses.

Done.

[Benedikt Meurer, 2014-02-17 13:22:39]

Committed patchset #2 manually as r19410 (presubmit successful).

[Toon Verwaest, 2014-02-18 08:47:09]

I find this change scary due to weak descriptor arrays. Can we change the api of
this function to take handles to their owning maps as input? That's the only way
we can guarantee that a large enough portion strays alive over GC.

[Benedikt Meurer, 2014-03-31 07:48:27]

Committed patchset #2 manually as r20339 (presubmit successful).