Reland of "Steps towards unification of number bitset and range types."

Reland of "Steps towards unification of number bitset and range types."

This reverts commit 76193749791eac9032f47db2f04efd78c0e120a9.

BUG=

Committed: https://crrev.com/5bd8407f8c8d7ed08574ca532ae5f1f4d6a8b39f
Cr-Commit-Position: refs/heads/master@{#26301}

[Jarin, 2015-01-27 07:13:44]

jarin@chromium.org changed reviewers:
+ bmeurer@chromium.org, rossberg@chromium.org

[Jarin, 2015-01-27 07:16:37]

Could you take a look, please?

I am introducing a hack to get to the isolate to allocate heap numbers, this
will be removed once we avoid using heap numbers in ranges.

[Benedikt Meurer, 2015-01-27 18:13:47]

So I guess that means that you just work-around the representation problem?

https://codereview.chromium.org/877643002/diff/40001/src/types.cc
File src/types.cc (right):

https://codereview.chromium.org/877643002/diff/40001/src/types.cc#newcode772
src/types.cc:772: i::Factory* f = i::Isolate::Current()->factory();
Urghs...this is awefull...but I guess we can live with that for a hopefully not
so long time.

[rossberg, 2015-01-27 19:13:20]

OK, why did the previous isolate hack not work?

Are the changes with respect to representations just temporary hacks as well?
Not sure I understand them.

[Jarin, 2015-01-28 08:39:37]

On 2015/01/27 19:13:20, rossberg wrote:
> OK, why did the previous isolate hack not work?
> 
> Are the changes with respect to representations just temporary hacks as well?
> Not sure I understand them.

Yes, the plan is to make the representations be really separate from the
semantic part.

The problem was that type Any with no representation bit sets is considered
inhabited, but in various places I ignored ranges with empty representations
(because I though they were uninhabited). The 'fix' just does not remove ranges
with empty representations as aggressively. I am still not entirely sure we will
always be fine, but all this is temporary until we separate the representations,
so I just want to make the tests happy.

As for the i::Isolate::Current() hack, this should go away really soon - I am
working on using unboxed doubles in ranges (well, at least in the zone).

[Benedikt Meurer, 2015-01-28 08:40:03]

The CQ bit was checked by bmeurer@chromium.org

[Benedikt Meurer, 2015-01-28 08:40:03]

lgtm

[commit-bot: I haz the power, 2015-01-28 08:40:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/877643002/40001

[commit-bot: I haz the power, 2015-01-28 08:42:23]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-01-28 08:42:44]

Patchset 3 (id:??) landed as
https://crrev.com/5bd8407f8c8d7ed08574ca532ae5f1f4d6a8b39f
Cr-Commit-Position: refs/heads/master@{#26301}