Simpler return value of HashTable::add/HashMap:add and others

Simpler return value of HashTable::add/HashMap:add and others

The current return value of HashTable::add() is a bool paired
with an iterator object. The creation and destruction of that
iterator object contributed to code size and since it was only
used at a few places it was not worth it.

Instead, let us return a bool paired with a pointer to the stored
object. That is after all what the code using the iterator object
wanted to have. I renamed the variable iterator -> storedValue
which is 99% [magic number] of the patch. Most users of HashTable
did not access iterator and did not have to change at all.

This change saves roughly 100 KB binary size of an x64 content_shell
built with clang. Other compilers, other architectures will save
different amounts, most likely less since most are not 64 bit.

The save is both in the instantiated ::add methods that don't have
to create an iterator object, and in the callers that don't have
to copy and destroy it.

R=morrita@chromium.org
RSR=eseidel@chromium.org

[Daniel Bratell, 2014-02-10 11:02:22]

This is the full patch that replaces the concept patch in
https://codereview.chromium.org/152883002/

Eric and Erik, could you have a look at your various parts and say if it looks
ok? Other people might want to check their code as well but most of the changes
are s/iterator/storedValue/ so not very exciting.

[Daniel Bratell, 2014-02-12 07:28:25]

eseidal, any opinion?

[Daniel Bratell, 2014-02-12 07:28:47]

On 2014/02/12 07:28:25, Daniel Bratell wrote:
> eseidal, any opinion?

eseidel*

[Daniel Bratell, 2014-02-13 16:25:46]

Anyone else with wtf ownership that can give their blessing. In the concept
patch (see the first message) eseidel liked the idea so the question is if the
end result is also good.

[gmorrita, 2014-02-13 18:38:53]

lgtm, given the what description says is promising.

[Daniel Bratell, 2014-02-13 20:17:42]

Thanks! I don't know how big the improvement will be on an mobile device, but I
see it as a thin slice off the top. Enough of them and the cheese is gone.

With the wtf changes reviewed, it's all the s/iterator/storedValue/ plus a few
~4 line changes that need lgtm too. I know it's boring, but it should be quick.

abarth, you want to boost your review stats?

[eseidel, 2014-02-13 21:37:41]

OK.  rslgtm.

[Daniel Bratell, 2014-02-14 08:26:59]

The CQ bit was checked by bratell@opera.com

[Daniel Bratell, 2014-02-14 15:46:33]

The CQ bit was unchecked by bratell@opera.com

[Daniel Bratell, 2014-02-14 15:46:52]

The CQ bit was checked by bratell@opera.com

[Daniel Bratell, 2014-02-14 17:00:07]

On 2014/02/14 15:46:52, Daniel Bratell wrote:
> The CQ bit was checked by mailto:bratell@opera.com

Missing Base URL so CQ can't do anything which it does without saying anything.
Opened https://codereview.chromium.org/167123002/ for CQ but need the lgtm and
rslgtm copied there.

[Daniel Bratell, 2014-02-14 17:00:11]

On 2014/02/14 15:46:52, Daniel Bratell wrote:
> The CQ bit was checked by mailto:bratell@opera.com

Missing Base URL so CQ can't do anything which it does without saying anything.
Opened https://codereview.chromium.org/167123002/ for CQ but need the lgtm and
rslgtm copied there.