Refactor strong mode to remove duplicate TypeRules

Refactor strong mode to remove duplicate TypeRules

Most of this code is already in StrongTypeSystemImpl. A few helpers move there, with some moving into CodeChecker. This does not rename "rules" parameters/variables yet, but let me know if we should do that.

This just moves code around. One subtle thing: TypeRules.isAssignableTo was only used in one place, in override checking, and that method just forwarded to TypeRules.isSubTypeOf. By contrast, StrongTypeSystemImpl does *not* just forward. So OverrideChecker now uses isSubtypeOf, preserving its old behavior.

R=leafp@google.com

Committed: https://github.com/dart-lang/sdk/commit/5e118574519d7277c725d9689b2f1a23c77b278c

[Jennifer Messerly, 2015-12-07 23:31:48]

Description was changed from

==========
Refactor strong mode to remove duplicate TypeRules
Most of this code is already in StrongTypeSystemImpl. A few helper smove there,
with some moving into CodeChecker
This does not rename "rules" parameters/variables yet, but let me know if we
should do that.
==========

to

==========
Refactor strong mode to remove duplicate TypeRules

Most of this code is already in StrongTypeSystemImpl. A few helpers move there,
with some moving into CodeChecker. This does not rename "rules"
parameters/variables yet, but let me know if we should do that.
==========

[Jennifer Messerly, 2015-12-07 23:31:48]

jmesserly@google.com changed reviewers:
+ leafp@google.com

[Jennifer Messerly, 2015-12-07 23:36:05]

Description was changed from

==========
Refactor strong mode to remove duplicate TypeRules

Most of this code is already in StrongTypeSystemImpl. A few helpers move there,
with some moving into CodeChecker. This does not rename "rules"
parameters/variables yet, but let me know if we should do that.
==========

to

==========
Refactor strong mode to remove duplicate TypeRules

Most of this code is already in StrongTypeSystemImpl. A few helpers move there,
with some moving into CodeChecker. This does not rename "rules"
parameters/variables yet, but let me know if we should do that.

This just moves code around. One subtle thing: TypeRules.isAssignableTo was only
used in one place, in override checking, and that method just forwarded to
TypeRules.isSubTypeOf. By contrast, StrongTypeSystemImpl does *not* just
forward. So OverrideChecker now uses isSubtypeOf, preserving its old behavior.
==========

[Leaf, 2015-12-08 00:24:40]

lgtm.  Feel free to leave the DownwardsInference thing for subsequent
refactoring.

https://codereview.chromium.org/1507933002/diff/40001/pkg/analyzer/lib/src/ta...
File pkg/analyzer/lib/src/task/strong/rules.dart (right):

https://codereview.chromium.org/1507933002/diff/40001/pkg/analyzer/lib/src/ta...
pkg/analyzer/lib/src/task/strong/rules.dart:17: class DownwardsInference {
This code is super close to dead.  I'm pretty sure that the only entry point is
inferExpression.  The annotateCast part is overridden in reify_coercions - I
think duping the inferExpression code into the checker and the reifier would be
a net win since you could get rid of this whole thing.

[Jennifer Messerly, 2015-12-08 01:06:32]

Thanks Leaf! PTAL

https://codereview.chromium.org/1507933002/diff/40001/pkg/analyzer/lib/src/ta...
File pkg/analyzer/lib/src/task/strong/rules.dart (right):

https://codereview.chromium.org/1507933002/diff/40001/pkg/analyzer/lib/src/ta...
pkg/analyzer/lib/src/task/strong/rules.dart:17: class DownwardsInference {
On 2015/12/08 00:24:40, Leaf wrote:
> This code is super close to dead.  I'm pretty sure that the only entry point
is
> inferExpression.  The annotateCast part is overridden in reify_coercions - I
> think duping the inferExpression code into the checker and the reifier would
be
> a net win since you could get rid of this whole thing.  

Ah yes! good catch. Done!

[Jennifer Messerly, 2015-12-08 01:07:11]

https://codereview.chromium.org/1507933002/diff/60001/pkg/analyzer/lib/src/ta...
File pkg/analyzer/lib/src/task/strong/checker.dart (right):

https://codereview.chromium.org/1507933002/diff/60001/pkg/analyzer/lib/src/ta...
pkg/analyzer/lib/src/task/strong/checker.dart:941: {cast: true}) {
I could remove this as well. WDYT?

[Leaf, 2015-12-08 01:12:11]

lgtm, thanks!

https://codereview.chromium.org/1507933002/diff/60001/pkg/analyzer/lib/src/ta...
File pkg/analyzer/lib/src/task/strong/checker.dart (right):

https://codereview.chromium.org/1507933002/diff/60001/pkg/analyzer/lib/src/ta...
pkg/analyzer/lib/src/task/strong/checker.dart:941: {cast: true}) {
On 2015/12/08 01:07:11, John Messerly wrote:
> I could remove this as well. WDYT?

sgtm

[Jennifer Messerly, 2015-12-08 01:19:17]

Description was changed from

==========
Refactor strong mode to remove duplicate TypeRules

Most of this code is already in StrongTypeSystemImpl. A few helpers move there,
with some moving into CodeChecker. This does not rename "rules"
parameters/variables yet, but let me know if we should do that.

This just moves code around. One subtle thing: TypeRules.isAssignableTo was only
used in one place, in override checking, and that method just forwarded to
TypeRules.isSubTypeOf. By contrast, StrongTypeSystemImpl does *not* just
forward. So OverrideChecker now uses isSubtypeOf, preserving its old behavior.
==========

to

==========
Refactor strong mode to remove duplicate TypeRules

Most of this code is already in StrongTypeSystemImpl. A few helpers move there,
with some moving into CodeChecker. This does not rename "rules"
parameters/variables yet, but let me know if we should do that.

This just moves code around. One subtle thing: TypeRules.isAssignableTo was only
used in one place, in override checking, and that method just forwarded to
TypeRules.isSubTypeOf. By contrast, StrongTypeSystemImpl does *not* just
forward. So OverrideChecker now uses isSubtypeOf, preserving its old behavior.

R=leafp@google.com

Committed:
https://github.com/dart-lang/sdk/commit/5e118574519d7277c725d9689b2f1a23c77b278c
==========

[Jennifer Messerly, 2015-12-08 01:19:18]

Committed patchset #5 (id:80001) manually as
5e118574519d7277c725d9689b2f1a23c77b278c (presubmit successful).