Refactor GetCodeCopyFromTemplate to get a single point where objects are replaced in code.

Refactor GetCodeCopyFromTemplate to get a single point where objects are replaced in code.

BUG=
R=verwaest@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=20035

[ulan, 2014-03-14 16:15:59]

PTAL

https://codereview.chromium.org/197283017/diff/1/src/ic.cc
File src/ic.cc (right):

https://codereview.chromium.org/197283017/diff/1/src/ic.cc#newcode1372
src/ic.cc:1372: isolate(), global, cell);
Passing raw pointers was a bug as GetCodeCopyFromTemplate allocates.

[Toon Verwaest, 2014-03-17 09:40:14]

Nice cleanup! Just one comment.

https://codereview.chromium.org/197283017/diff/1/src/ic.cc
File src/ic.cc (right):

https://codereview.chromium.org/197283017/diff/1/src/ic.cc#newcode1372
src/ic.cc:1372: isolate(), global, cell);
Nice catch!

On 2014/03/14 16:15:59, ulan wrote:
> Passing raw pointers was a bug as GetCodeCopyFromTemplate allocates.

https://codereview.chromium.org/197283017/diff/1/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/197283017/diff/1/src/objects.cc#newcode10521
src/objects.cc:10521: }
Why don't we just introduce an invariant that the objects in the pattern are
required to be in the right order? That way we don't have this O(m*n) algo, but
just O(m).

Then add a second invariant that all objects in the pattern are required to be
used.

[ulan, 2014-03-17 13:43:31]

https://codereview.chromium.org/197283017/diff/1/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/197283017/diff/1/src/objects.cc#newcode10521
src/objects.cc:10521: }
On 2014/03/17 09:40:14, Toon Verwaest wrote:
> Why don't we just introduce an invariant that the objects in the pattern are
> required to be in the right order? That way we don't have this O(m*n) algo,
but
> just O(m).
> 
> Then add a second invariant that all objects in the pattern are required to be
> used.
Good idea, turned that into O(m) and added comment explaining the order. I hope
the UNREACHABLE below will catch possible future misuses.

[Toon Verwaest, 2014-03-17 14:04:23]

lgtm

[ulan, 2014-03-18 14:15:19]

Committed patchset #2 manually as r20035 (presubmit successful).