dart:io | Add asynchronous versions of the methods of FileSystemEntity and Link.

dart:io | Add asynchronous versions of the methods of FileSystemEntity and Link.

BUG=
R=sgjesse@google.com

Committed: https://code.google.com/p/dart/source/detail?r=22354

[Bill Hesse, 2013-05-02 14:02:42]

https://codereview.chromium.org/14642012/diff/2001/sdk/lib/io/file_system_ent...
File sdk/lib/io/file_system_entity.dart (right):

https://codereview.chromium.org/14642012/diff/2001/sdk/lib/io/file_system_ent...
sdk/lib/io/file_system_entity.dart:61: _throwIfError(result, 'Error in
FileSystemEntity.identical');
Error should be on the future.

https://codereview.chromium.org/14642012/diff/2001/sdk/lib/io/file_system_ent...
sdk/lib/io/file_system_entity.dart:61: _throwIfError(result, 'Error in
FileSystemEntity.identical');
Fixed - all errors are now on the Futures.

https://codereview.chromium.org/14642012/diff/8001/sdk/lib/io/file_system_ent...
File sdk/lib/io/file_system_entity.dart (right):

https://codereview.chromium.org/14642012/diff/8001/sdk/lib/io/file_system_ent...
sdk/lib/io/file_system_entity.dart:80: => new Future<FileSystemEntityType>(
This function could be changed to its final, correct form using _getTypeAsync
already, as the ones below are.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
File tests/standalone/io/file_system_async_links_test.dart (right):

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:33: .then((link) =>
Expect.equals(y, link.path))
The indentation has tried to follow the rule "Follow the rules, but a closure or
a .then does not need to be indented"

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:34: .then((_) =>
FutureExpect.isFalse(new File(y).exists()))
Note that this also could be written
.then((_) => new File(y).exists()).then(Expect.isFalse)

But to make the changes uniform on all these files, since .equals, .equalsList,
and .throws cannot be done this way, we use FutureExpect.

[Søren Gjesse, 2013-05-03 07:13:36]

LGTM

I really like the comprehensive testing!

https://codereview.chromium.org/14642012/diff/8001/sdk/lib/io/file_system_ent...
File sdk/lib/io/file_system_entity.dart (right):

https://codereview.chromium.org/14642012/diff/8001/sdk/lib/io/file_system_ent...
sdk/lib/io/file_system_entity.dart:47: new Future(() => _getTypeSync(path,
followLinks));
This still needs to be calling a service finction on the native thread to be
truly async.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
File tests/standalone/io/file_system_async_links_test.dart (right):

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:28: Future
testFileExistsCreate() =>
Add a receive port to keep the test alive until the lsat async operation
completes.

For the other tests as well.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:28: Future
testFileExistsCreate() =>
Normally we don't use the => syntax for long methods.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:49: .then((_) => new
File(y).create())
Shouldn't this be creating the file x - or am I missing something here? Up til
this point the only file system object which exists is the link y -> x.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:172:
testDirectoryListing()  =>
Please add explicit return type

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/standalon...
File tests/standalone/standalone.status (right):

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/standalon...
tests/standalone/standalone.status:51: # Links on Windows are tested by
windows_file_system_[async_]links_test instead.
How about moving this comment into the files and just return from main on
Windows? This keeps the status file cleaner.

[Bill Hesse, 2013-05-03 08:39:31]

Committed patchset #5 manually as r22354 (presubmit successful).

[Bill Hesse, 2013-05-03 08:44:06]

https://codereview.chromium.org/14642012/diff/8001/sdk/lib/io/file_system_ent...
File sdk/lib/io/file_system_entity.dart (right):

https://codereview.chromium.org/14642012/diff/8001/sdk/lib/io/file_system_ent...
sdk/lib/io/file_system_entity.dart:47: new Future(() => _getTypeSync(path,
followLinks));
On 2013/05/03 07:13:36, Søren Gjesse wrote:
> This still needs to be calling a service finction on the native thread to be
> truly async.

This is done in a follow-up CL, already written.  I can merge them, but I prefer
not to.

https://codereview.chromium.org/14642012/diff/8001/sdk/lib/io/file_system_ent...
sdk/lib/io/file_system_entity.dart:80: => new Future<FileSystemEntityType>(
On 2013/05/02 14:02:42, Bill Hesse wrote:
> This function could be changed to its final, correct form using _getTypeAsync
> already, as the ones below are.

Done in the second CL.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
File tests/standalone/io/file_system_async_links_test.dart (right):

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:28: Future
testFileExistsCreate() =>
On 2013/05/03 07:13:36, Søren Gjesse wrote:
> Add a receive port to keep the test alive until the lsat async operation
> completes.
> 
> For the other tests as well.

This is done in the main method, which calls this.
The tests can be run serially, or in parallel, since they all return Futures. 
Right now they are run serially, but perhaps we should shift to parallel.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:28: Future
testFileExistsCreate() =>
On 2013/05/03 07:13:36, Søren Gjesse wrote:
> Normally we don't use the => syntax for long methods.

In this case, it was possible for all of them, since they all are a single
expression.  But I will change them.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/io/file_s...
tests/standalone/io/file_system_async_links_test.dart:49: .then((_) => new
File(y).create())
On 2013/05/03 07:13:36, Søren Gjesse wrote:
> Shouldn't this be creating the file x - or am I missing something here? Up til
> this point the only file system object which exists is the link y -> x.

This is testing that file creation through a link succeeds - that a broken link
with a nonexistent target creates its target if create is called with the link
path.  That is the semantics (on posix), and is what is tested here.

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/standalon...
File tests/standalone/standalone.status (right):

https://codereview.chromium.org/14642012/diff/8001/tests/standalone/standalon...
tests/standalone/standalone.status:51: # Links on Windows are tested by
windows_file_system_[async_]links_test instead.
On 2013/05/03 07:13:36, Søren Gjesse wrote:
> How about moving this comment into the files and just return from main on
> Windows? This keeps the status file cleaner.

Done.  Also added to file_read_special_device_test, so the entire section is
removed.  Comment added to both posix and windows tests, mentioning the other.