Start migrating error message templates to the runtime.

Start migrating error message templates to the runtime.

Currently done with two templates, one used from native js, one from runtime.

R=verwaest@chromium.org

Committed: https://crrev.com/a5ac0290588f44dfccba571393761747c48e0e7b
Cr-Commit-Position: refs/heads/master@{#27864}

[Yang, 2015-04-15 13:33:42]

On 2015/04/14 13:43:52, Yang wrote:

ping

[Toon Verwaest, 2015-04-15 15:02:46]

lgtm

[adamk, 2015-04-15 17:30:36]

adamk@chromium.org changed reviewers:
+ adamk@chromium.org

[adamk, 2015-04-15 17:30:36]

Can you give some background on this change? (A more detailed CL description
would've helped.)

[arv (Not doing code reviews), 2015-04-15 17:58:38]

On 2015/04/15 17:30:36, adamk wrote:
> Can you give some background on this change? (A more detailed CL description
> would've helped.)

I'm also curious about this. We've earlier talked about extending the mjsunit
tests to verify the message and this will make that more problematic.

If this is for perf than that trumps any of that of course.

[Yang, 2015-04-15 18:34:03]

On 2015/04/15 17:58:38, arv wrote:
> On 2015/04/15 17:30:36, adamk wrote:
> > Can you give some background on this change? (A more detailed CL description
> > would've helped.)
> 
> I'm also curious about this. We've earlier talked about extending the mjsunit
> tests to verify the message and this will make that more problematic.
> 
> If this is for perf than that trumps any of that of course.

The motivation is to reduce the memory footprint per context. Many pages contain
many iframes, each of which has its own context. Each context has a copy of this
message template dictionary, which takes significant space on the heap, even
though the string literals are shared across the whole isolate. By removing that
message template dictionary, we save 13kB (out of 205kB) per context and 28kB
(out of 827kB) per isolate, on x64.

[arv (Not doing code reviews), 2015-04-15 19:50:40]

On 2015/04/15 18:34:03, Yang wrote:
> On 2015/04/15 17:58:38, arv wrote:
> > On 2015/04/15 17:30:36, adamk wrote:
> > > Can you give some background on this change? (A more detailed CL
description
> > > would've helped.)
> > 
> > I'm also curious about this. We've earlier talked about extending the
mjsunit
> > tests to verify the message and this will make that more problematic.
> > 
> > If this is for perf than that trumps any of that of course.
> 
> The motivation is to reduce the memory footprint per context. Many pages
contain
> many iframes, each of which has its own context. Each context has a copy of
this
> message template dictionary, which takes significant space on the heap, even
> though the string literals are shared across the whole isolate. By removing
that
> message template dictionary, we save 13kB (out of 205kB) per context and 28kB
> (out of 827kB) per isolate, on x64.

LGTM

That is a big win.

I wonder if this is going to be the trend from now on then? If so maybe we
should find a way to share a context between frames that can contain
implementations like these?

[Yang, 2015-04-16 04:43:50]

The CQ bit was checked by yangguo@chromium.org

[commit-bot: I haz the power, 2015-04-16 04:43:57]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1087633005/1

[commit-bot: I haz the power, 2015-04-16 04:57:23]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-04-16 04:57:24]

Try jobs failed on following builders:
  v8_win_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_rel/builds/4976)

[Yang, 2015-04-16 06:09:36]

The CQ bit was checked by yangguo@chromium.org

[Yang, 2015-04-16 06:09:36]

The patchset sent to the CQ was uploaded after l-g-t-m from arv@chromium.org,
verwaest@chromium.org
Link to the patchset: https://codereview.chromium.org/1087633005/#ps20001
(title: "rebase")

[commit-bot: I haz the power, 2015-04-16 06:09:46]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1087633005/20001

[commit-bot: I haz the power, 2015-04-16 06:25:52]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-04-16 06:25:53]

Try jobs failed on following builders:
  v8_win_nosnap_shared_compile_rel on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_win_nosnap_shared_compil...)

[Yang, 2015-04-16 06:36:54]

The CQ bit was checked by yangguo@chromium.org

[Yang, 2015-04-16 06:36:54]

The patchset sent to the CQ was uploaded after l-g-t-m from arv@chromium.org,
verwaest@chromium.org
Link to the patchset: https://codereview.chromium.org/1087633005/#ps40001
(title: "fix windows build")

[commit-bot: I haz the power, 2015-04-16 06:36:58]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1087633005/40001

[commit-bot: I haz the power, 2015-04-16 07:01:25]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-04-16 07:01:45]

Patchset 3 (id:??) landed as
https://crrev.com/a5ac0290588f44dfccba571393761747c48e0e7b
Cr-Commit-Position: refs/heads/master@{#27864}