Handle deferred access as pre-step in SemanticSendVisitor.

Handle deferred access as pre-step in SemanticSendVisitor.

BUG=
R=floitsch@google.com, sigurdm@google.com

Committed: https://github.com/dart-lang/sdk/commit/6f92dfe8ff059e19de5e349545bcb6775d6ae9fa

[Johnni Winther, 2015-07-16 11:56:16]

johnniwinther@google.com changed reviewers:
+ floitsch@google.com, sigurdm@google.com

[floitsch, 2015-07-16 15:24:44]

LGTM.

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
File pkg/compiler/lib/src/resolution/members.dart (right):

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
pkg/compiler/lib/src/resolution/members.dart:334: } else if (element.isClass &&
sendIsMemberAccess){
missing space before "{".

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
pkg/compiler/lib/src/resolution/members.dart:1964: result = new
PrefixResult(prefix, result.element);
If I understand correctly, this is the case where we have `prefix.Class.foo`,
and the node is "prefix.Class".

Add comment, that the `prefix` might be deferred, but that we don't need to run
through 'handleDeferredAccess', because that's only necessary for the actual
access (which would be done in [handlePrefixSend] below.

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
File pkg/compiler/lib/src/resolution/send_structure.dart (right):

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
pkg/compiler/lib/src/resolution/send_structure.dart:2314: /// The send structure
of for the whole [Send] node. For instance a
-of-

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolv...
File pkg/compiler/lib/src/resolved_visitor.dart (right):

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolv...
pkg/compiler/lib/src/resolved_visitor.dart:392: void visitDeferredAccess(
I wonder if we want to rename the "visit" into something else, since it doesn't
follow the other's visit pattern.
I don't have a good name, though.

[sigurdm, 2015-07-20 07:52:21]

LGTM

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/ssa/bu...
File pkg/compiler/lib/src/ssa/builder.dart (right):

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/ssa/bu...
pkg/compiler/lib/src/ssa/builder.dart:3422:
generateIsDeferredLoadedCheckIfNeeded(prefix, node);
Remove the check from all covered cases.

[Johnni Winther, 2015-07-20 08:53:53]

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
File pkg/compiler/lib/src/resolution/members.dart (right):

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
pkg/compiler/lib/src/resolution/members.dart:334: } else if (element.isClass &&
sendIsMemberAccess){
On 2015/07/16 15:24:44, floitsch wrote:
> missing space before "{".

Done.

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
pkg/compiler/lib/src/resolution/members.dart:1964: result = new
PrefixResult(prefix, result.element);
On 2015/07/16 15:24:44, floitsch wrote:
> If I understand correctly, this is the case where we have `prefix.Class.foo`,
> and the node is "prefix.Class".
> 
> Add comment, that the `prefix` might be deferred, but that we don't need to
run
> through 'handleDeferredAccess', because that's only necessary for the actual
> access (which would be done in [handlePrefixSend] below.

Done.

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
File pkg/compiler/lib/src/resolution/send_structure.dart (right):

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolu...
pkg/compiler/lib/src/resolution/send_structure.dart:2314: /// The send structure
of for the whole [Send] node. For instance a
On 2015/07/16 15:24:44, floitsch wrote:
> -of-

Done.

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolv...
File pkg/compiler/lib/src/resolved_visitor.dart (right):

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/resolv...
pkg/compiler/lib/src/resolved_visitor.dart:392: void visitDeferredAccess(
On 2015/07/16 15:24:44, floitsch wrote:
> I wonder if we want to rename the "visit" into something else, since it
doesn't
> follow the other's visit pattern.
> I don't have a good name, though.

We'll try 'previsitDeferredAccess'

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/ssa/bu...
File pkg/compiler/lib/src/ssa/builder.dart (right):

https://codereview.chromium.org/1238783003/diff/1/pkg/compiler/lib/src/ssa/bu...
pkg/compiler/lib/src/ssa/builder.dart:3422:
generateIsDeferredLoadedCheckIfNeeded(prefix, node);
On 2015/07/20 07:52:21, sigurdm wrote:
> Remove the check from all covered cases.

Done.

[Johnni Winther, 2015-07-20 09:18:33]

Committed patchset #2 (id:20001) manually as
6f92dfe8ff059e19de5e349545bcb6775d6ae9fa (presubmit successful).