Encode super calls via extra properties on prototypes.

Encode super calls via extra properties on prototypes.

BUG= http://dartbug.com/21666
R=ahe@google.com, floitsch@google.com

Committed: https://code.google.com/p/dart/source/detail?r=41989

[herhut, 2014-11-24 10:01:02]

herhut@google.com changed reviewers:
+ ahe@google.com

[herhut, 2014-11-24 10:06:19]

This is essentially the same code as before except for changes in
js_mirrors.dart. I have added code to filter out stubs (which should take care
of the order dependency that was there before) and to filter out aliases. 

For the latter, I have decided to put the aliased name onto the function object,
which allows me to identify aliases by checking for that property and comparing
with the name. Compared to the array solution we discussed, I do not need to
store another piece of meta data and I do not need to search through it.

[floitsch, 2014-11-24 10:47:51]

floitsch@google.com changed reviewers:
+ floitsch@google.com

[floitsch, 2014-11-24 10:47:53]

DBC.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
File pkg/compiler/lib/src/js_backend/backend.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
pkg/compiler/lib/src/js_backend/backend.dart:335: * A set of members that are
called from subclasses via super.
`super`

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
pkg/compiler/lib/src/js_backend/backend.dart:565: * Returns `true` is [member]
is called from a subclass via `super`.
.. `true` if [member] ...

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
File pkg/compiler/lib/src/js_backend/minify_namer.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
pkg/compiler/lib/src/js_backend/minify_namer.dart:21: final String superPrefix =
r's$';
Are you sure that you cannot have a conflict with a setter?

ex: setterPrefix + "$a" == superPrefix + "a"

Edit:
Looks like the superPrefix is just used as a suggestion. So should be fine. But
please confirm.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
File pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart:434: name =
"$name:${namer.getNameOfAliasedSuperMember(member)}";
Too far away from the code that extracts it.
Try to move them closer together.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart:445:
builder.addProperty(namer.invocationName(selector), function));
long line.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
File pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart:543: if (#)
splitOffAliases(constructor);
Please change this to named js-holes. (you will need to change all holes in this
template).
See line 505 for an example.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
File pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart
(right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart:150: var
name = nameTag.split(':')[0];
Ideally this code would also be together with the one that encodes it.
I guess the only easy way is to use string-interpolation to do that.

[ahe, 2014-11-25 09:33:03]

I think there is a real problem using "s$" in minified mode.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
File pkg/compiler/lib/src/js_backend/minify_namer.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
pkg/compiler/lib/src/js_backend/minify_namer.dart:21: final String superPrefix =
r's$';
On 2014/11/24 10:47:52, floitsch wrote:
> Are you sure that you cannot have a conflict with a setter?
> 
> ex: setterPrefix + "$a" == superPrefix + "a"
> 
> Edit:
> Looks like the superPrefix is just used as a suggestion. So should be fine.
But
> please confirm.

I think this is problematic. It will likely confuse compact noSuchMethod
handling, see 
pkg/compiler/lib/src/js_emitter/old_emitter/nsm_emitter.dart line 362.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
File pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart:434: name =
"$name:${namer.getNameOfAliasedSuperMember(member)}";
I'm concerned that using ':' (colon) here is likely to create hard to debug
bugs. The mirrors library already uses colons in names.

I suggest that you add this information to the "structured info" array computed
below.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
File pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart:363: List
buildSplitOffAliases() {
If you follow my suggestion in container_builder.dart, then this code can be
removed.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart:543: if (#)
splitOffAliases(constructor);
On 2014/11/24 10:47:52, floitsch wrote:
> Please change this to named js-holes. (you will need to change all holes in
this
> template).
> See line 505 for an example.

Awesome. I needed that :-)

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
File pkg/compiler/lib/src/ssa/codegen.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
pkg/compiler/lib/src/ssa/codegen.dart:1685: push(
I assume this is because super getters aren't supported yet. I don't think it is
too hard to extend the tearOff code to handle this situation.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
pkg/compiler/lib/src/ssa/codegen.dart:1694: use(node.inputs.first);
Is there a more descriptive name for this node?

https://codereview.chromium.org/753113002/diff/1/sdk/lib/_internal/compiler/j...
File sdk/lib/_internal/compiler/js_lib/js_mirrors.dart (right):

https://codereview.chromium.org/753113002/diff/1/sdk/lib/_internal/compiler/j...
sdk/lib/_internal/compiler/js_lib/js_mirrors.dart:737: if (JS('bool',
r'#.$getterStub', method._jsFunction)) continue;
Need to use r'!!#.$getterStub' to ensure that the expression really returns a
bool.

https://codereview.chromium.org/753113002/diff/1/sdk/lib/_internal/compiler/j...
sdk/lib/_internal/compiler/js_lib/js_mirrors.dart:2913: bool
isAliasedSuperMethod(var jsFunction, String key) {
I can imagine wanting more than one alias for a function, so consider storing
its real name instead.

https://codereview.chromium.org/753113002/diff/1/sdk/lib/_internal/compiler/j...
sdk/lib/_internal/compiler/js_lib/js_mirrors.dart:2914: var aliasName =
JS('String', r'#.$aliasName', jsFunction);
I think this has to be:

JS('String|Null', r'#.$aliasName', jsFunction)

[herhut, 2014-11-25 14:03:20]

PTAL

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
File pkg/compiler/lib/src/js_backend/backend.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
pkg/compiler/lib/src/js_backend/backend.dart:335: * A set of members that are
called from subclasses via super.
On 2014/11/24 10:47:52, floitsch wrote:
> `super`

Done.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
pkg/compiler/lib/src/js_backend/backend.dart:565: * Returns `true` is [member]
is called from a subclass via `super`.
On 2014/11/24 10:47:52, floitsch wrote:
> .. `true` if [member] ...

Done.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
File pkg/compiler/lib/src/js_backend/minify_namer.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
pkg/compiler/lib/src/js_backend/minify_namer.dart:21: final String superPrefix =
r's$';
On 2014/11/24 10:47:52, floitsch wrote:
> Are you sure that you cannot have a conflict with a setter?
> 
> ex: setterPrefix + "$a" == superPrefix + "a"
> 
> Edit:
> Looks like the superPrefix is just used as a suggestion. So should be fine.
But
> please confirm.

We cannot have a $a, as it is not a valid name. But regardless, the prefix
really is just used to generate a different hash. There even is no need to use a
different one for minified code anymore. I have removed it.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_back...
pkg/compiler/lib/src/js_backend/minify_namer.dart:21: final String superPrefix =
r's$';
On 2014/11/25 09:33:03, ahe wrote:
> On 2014/11/24 10:47:52, floitsch wrote:
> > Are you sure that you cannot have a conflict with a setter?
> > 
> > ex: setterPrefix + "$a" == superPrefix + "a"
> > 
> > Edit:
> > Looks like the superPrefix is just used as a suggestion. So should be fine.
> But
> > please confirm.
> 
> I think this is problematic. It will likely confuse compact noSuchMethod
> handling, see 
> pkg/compiler/lib/src/js_emitter/old_emitter/nsm_emitter.dart line 362.

Acknowledged.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
File pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart:434: name =
"$name:${namer.getNameOfAliasedSuperMember(member)}";
On 2014/11/25 09:33:03, ahe wrote:
> I'm concerned that using ':' (colon) here is likely to create hard to debug
> bugs. The mirrors library already uses colons in names.
> 
> I suggest that you add this information to the "structured info" array
computed
> below.

Done.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart:434: name =
"$name:${namer.getNameOfAliasedSuperMember(member)}";
On 2014/11/24 10:47:52, floitsch wrote:
> Too far away from the code that extracts it.
> Try to move them closer together.

This has been removed.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/container_builder.dart:445:
builder.addProperty(namer.invocationName(selector), function));
On 2014/11/24 10:47:52, floitsch wrote:
> long line.

Done.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
File pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart:363: List
buildSplitOffAliases() {
On 2014/11/25 09:33:03, ahe wrote:
> If you follow my suggestion in container_builder.dart, then this code can be
> removed.

Done.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart:543: if (#)
splitOffAliases(constructor);
On 2014/11/24 10:47:52, floitsch wrote:
> Please change this to named js-holes. (you will need to change all holes in
this
> template).
> See line 505 for an example.

Done.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart:543: if (#)
splitOffAliases(constructor);
On 2014/11/25 09:33:03, ahe wrote:
> On 2014/11/24 10:47:52, floitsch wrote:
> > Please change this to named js-holes. (you will need to change all holes in
> this
> > template).
> > See line 505 for an example.
> 
> Awesome. I needed that :-)

Acknowledged.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
File pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart
(right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/js_emit...
pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart:150: var
name = nameTag.split(':')[0];
On 2014/11/24 10:47:52, floitsch wrote:
> Ideally this code would also be together with the one that encodes it.
> I guess the only easy way is to use string-interpolation to do that.

Removed.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
File pkg/compiler/lib/src/ssa/codegen.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
pkg/compiler/lib/src/ssa/codegen.dart:1685: push(
On 2014/11/25 09:33:03, ahe wrote:
> I assume this is because super getters aren't supported yet. I don't think it
is
> too hard to extend the tearOff code to handle this situation.

It simply was not common enough to optimize for.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
pkg/compiler/lib/src/ssa/codegen.dart:1694: use(node.inputs.first);
On 2014/11/25 09:33:03, ahe wrote:
> Is there a more descriptive name for this node?

The entire file uses [node] for the current node being processed. What it is
follows from the method you're in. While it can debated whether that is a good
idea or not, at least it currently is consistent.

https://codereview.chromium.org/753113002/diff/1/sdk/lib/_internal/compiler/j...
File sdk/lib/_internal/compiler/js_lib/js_mirrors.dart (right):

https://codereview.chromium.org/753113002/diff/1/sdk/lib/_internal/compiler/j...
sdk/lib/_internal/compiler/js_lib/js_mirrors.dart:737: if (JS('bool',
r'#.$getterStub', method._jsFunction)) continue;
On 2014/11/25 09:33:03, ahe wrote:
> Need to use r'!!#.$getterStub' to ensure that the expression really returns a
> bool.

Done.

https://codereview.chromium.org/753113002/diff/1/sdk/lib/_internal/compiler/j...
sdk/lib/_internal/compiler/js_lib/js_mirrors.dart:2913: bool
isAliasedSuperMethod(var jsFunction, String key) {
On 2014/11/25 09:33:03, ahe wrote:
> I can imagine wanting more than one alias for a function, so consider storing
> its real name instead.

Done.

https://codereview.chromium.org/753113002/diff/1/sdk/lib/_internal/compiler/j...
sdk/lib/_internal/compiler/js_lib/js_mirrors.dart:2914: var aliasName =
JS('String', r'#.$aliasName', jsFunction);
On 2014/11/25 09:33:03, ahe wrote:
> I think this has to be:
> 
> JS('String|Null', r'#.$aliasName', jsFunction)

Done.

[ahe, 2014-11-25 14:10:33]

Replying to comments. Will take a closer look later.

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
File pkg/compiler/lib/src/ssa/codegen.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
pkg/compiler/lib/src/ssa/codegen.dart:1694: use(node.inputs.first);
On 2014/11/25 14:03:20, herhut wrote:
> On 2014/11/25 09:33:03, ahe wrote:
> > Is there a more descriptive name for this node?
> 
> The entire file uses [node] for the current node being processed. What it is
> follows from the method you're in. While it can debated whether that is a good
> idea or not, at least it currently is consistent.

What I meant is: Is there a more descriptive name for this expression:
"node.inputs.first". Often there is a getter with a more descriptive name.

If adding a getter isn't a great idea, then perhaps storing the node in a
variable with a descriptive name would help.

HInstruction descriptiveName = node.inputs.first;

[herhut, 2014-11-25 14:25:01]

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
File pkg/compiler/lib/src/ssa/codegen.dart (right):

https://codereview.chromium.org/753113002/diff/1/pkg/compiler/lib/src/ssa/cod...
pkg/compiler/lib/src/ssa/codegen.dart:1694: use(node.inputs.first);
On 2014/11/25 14:10:33, ahe wrote:
> On 2014/11/25 14:03:20, herhut wrote:
> > On 2014/11/25 09:33:03, ahe wrote:
> > > Is there a more descriptive name for this node?
> > 
> > The entire file uses [node] for the current node being processed. What it is
> > follows from the method you're in. While it can debated whether that is a
good
> > idea or not, at least it currently is consistent.
> 
> What I meant is: Is there a more descriptive name for this expression:
> "node.inputs.first". Often there is a getter with a more descriptive name.
> 
> If adding a getter isn't a great idea, then perhaps storing the node in a
> variable with a descriptive name would help.
> 
> HInstruction descriptiveName = node.inputs.first;

Got you, there is. Done.

[floitsch, 2014-11-25 14:25:13]

Still LGTM.

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
File pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart (right):

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart:693:
..add(js('$finishClassesName = #', finishClassesFunction))
fix indentation (already previous line).

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
File pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart
(right):

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart:147: f =
typeof alias == "string" ? array[++index]
Use an if. The "?" gets too complicated.

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart:151:
descriptor[name] = descriptor[alias] = f];
indentation.

[sra1, 2014-11-25 18:02:48]

sra@google.com changed reviewers:
+ sra@google.com

[sra1, 2014-11-25 18:02:48]

https://codereview.chromium.org/753113002/diff/80001/pkg/compiler/lib/src/js_...
File pkg/compiler/lib/src/js_backend/namer.dart (right):

https://codereview.chromium.org/753113002/diff/80001/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_backend/namer.dart:818: return
getMappedInstanceName(proposal);
Maybe add a TODO.
Since super aliases are always known to exist there is no requirement like that
of a dynamic call to trigger noSuchMethod.  This means super aliases could be
assigned more compactly, more like fields.

[herhut, 2014-11-26 09:41:36]

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
File pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart (right):

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_emitter/old_emitter/emitter.dart:693:
..add(js('$finishClassesName = #', finishClassesFunction))
On 2014/11/25 14:25:13, floitsch wrote:
> fix indentation (already previous line).

Done.

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
File pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart
(right):

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart:147: f =
typeof alias == "string" ? array[++index]
On 2014/11/25 14:25:13, floitsch wrote:
> Use an if. The "?" gets too complicated.

Done.

https://codereview.chromium.org/753113002/diff/40001/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart:151:
descriptor[name] = descriptor[alias] = f];
On 2014/11/25 14:25:13, floitsch wrote:
> indentation.

Done.

https://codereview.chromium.org/753113002/diff/80001/pkg/compiler/lib/src/js_...
File pkg/compiler/lib/src/js_backend/namer.dart (right):

https://codereview.chromium.org/753113002/diff/80001/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_backend/namer.dart:818: return
getMappedInstanceName(proposal);
On 2014/11/25 18:02:48, sra1 wrote:
> Maybe add a TODO.
> Since super aliases are always known to exist there is no requirement like
that
> of a dynamic call to trigger noSuchMethod.  This means super aliases could be
> assigned more compactly, more like fields.

Thanks! Great catch! Done.

[ahe, 2014-11-26 10:45:43]

LGTM

https://codereview.chromium.org/753113002/diff/20002/pkg/compiler/lib/src/js_...
File pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart
(right):

https://codereview.chromium.org/753113002/diff/20002/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart:154:
descriptor[alias] = f];
Excellent. I'm just now producing a CL that calls addStubs and this eliminates
one special case.

I still need to carefully handle when a new super call has been added, as well
as remove the alias if a method has been removed.

https://codereview.chromium.org/753113002/diff/20002/pkg/compiler/lib/src/ssa...
File pkg/compiler/lib/src/ssa/codegen.dart (right):

https://codereview.chromium.org/753113002/diff/20002/pkg/compiler/lib/src/ssa...
pkg/compiler/lib/src/ssa/codegen.dart:1698: visitArguments(node.inputs, start:
1)]),
Perhaps add a comment to the start parameter: skipping receiver.

[herhut, 2014-11-26 11:35:59]

https://codereview.chromium.org/753113002/diff/20002/pkg/compiler/lib/src/js_...
File pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart
(right):

https://codereview.chromium.org/753113002/diff/20002/pkg/compiler/lib/src/js_...
pkg/compiler/lib/src/js_emitter/old_emitter/reflection_data_parser.dart:154:
descriptor[alias] = f];
On 2014/11/26 10:45:43, ahe wrote:
> Excellent. I'm just now producing a CL that calls addStubs and this eliminates
> one special case.
> 
> I still need to carefully handle when a new super call has been added, as well
> as remove the alias if a method has been removed.

Acknowledged.

https://codereview.chromium.org/753113002/diff/20002/pkg/compiler/lib/src/ssa...
File pkg/compiler/lib/src/ssa/codegen.dart (right):

https://codereview.chromium.org/753113002/diff/20002/pkg/compiler/lib/src/ssa...
pkg/compiler/lib/src/ssa/codegen.dart:1698: visitArguments(node.inputs, start:
1)]),
On 2014/11/26 10:45:43, ahe wrote:
> Perhaps add a comment to the start parameter: skipping receiver.

Done.

[herhut, 2014-11-26 11:48:44]

Committed patchset #7 (id:110001) manually as 41989 (presubmit successful).