SocketDescriptor should not be used as a boolean

SocketDescriptor should not be used as a boolean

In UnixDomainServerSocketTest.AcceptWithForbiddenUser, a
SocketDescriptor value was being tested as a boolean, which
shouldn't happen. Instead, compare to kInvalidSocket.

Also, build trivial external_application_unittests on Windows.

BUG=None
TEST=mojob.py test
R=jam@chromium.org, davemoore, jam

Committed: https://chromium.googlesource.com/external/mojo/+/b8ee9fe1beee80d0bb0ee9bd24e78dba917597d2

[Chris Masone, 2014-11-17 18:10:56]

PTAL. This fixes the currently failing tests.

[jam, 2014-11-17 19:01:26]

lgtm

[Chris Masone, 2014-11-17 19:17:21]

Committed patchset #1 (id:1) manually as
b8ee9fe1beee80d0bb0ee9bd24e78dba917597d2 (presubmit successful).

[jamesr, 2014-11-17 21:19:14]

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

[jamesr, 2014-11-17 21:19:15]

This doesn't build either:

[2542/5587] CXX
obj/mojo/services/public/cpp/view_manager/lib/view_manager.view_manager_client_factory.obj
FAILED: E:/b/depot_tools/python276_bin/python.exe gyp-win-tool link-wrapper
environment.x64 False link.exe /nologo /OUT:external_application_unittests.exe
/PDB:external_application_unittests.exe.pdb
@external_application_unittests.exe.rsp
external_application_unittests.external_application_test_main.obj : error
LNK2019: unresolved external symbol "void __cdecl mojo::embedder::Init(class
scoped_ptr<class mojo::embedder::PlatformSupport,struct
base::DefaultDeleter<class mojo::embedder::PlatformSupport> >)"
(?Init@embedder@mojo@@YAXV?$scoped_ptr@VPlatformSupport@embedder@mojo@@U?$DefaultDeleter@VPlatformSupport@embedder@mojo@@@base@@@@@Z)
referenced in function main


does it make any sense to build this on windows?  The whole external application
connection mechanism is based on POSIX domain sockets and we have no use case
for this on windows (afaik)

[Chris Masone, 2014-11-17 21:21:23]

On 2014/11/17 21:19:15, jamesr wrote:
> This doesn't build either:
> 
> [2542/5587] CXX
>
obj/mojo/services/public/cpp/view_manager/lib/view_manager.view_manager_client_factory.obj
> FAILED: E:/b/depot_tools/python276_bin/python.exe gyp-win-tool link-wrapper
> environment.x64 False link.exe /nologo /OUT:external_application_unittests.exe
> /PDB:external_application_unittests.exe.pdb
> @external_application_unittests.exe.rsp
> external_application_unittests.external_application_test_main.obj : error
> LNK2019: unresolved external symbol "void __cdecl mojo::embedder::Init(class
> scoped_ptr<class mojo::embedder::PlatformSupport,struct
> base::DefaultDeleter<class mojo::embedder::PlatformSupport> >)"
>
(?Init@embedder@mojo@@YAXV?$scoped_ptr@VPlatformSupport@embedder@mojo@@U?$DefaultDeleter@VPlatformSupport@embedder@mojo@@@base@@@@@Z)
> referenced in function main
> 
> 
> does it make any sense to build this on windows?  The whole external
application
> connection mechanism is based on POSIX domain sockets and we have no use case
> for this on windows (afaik)

It doesn't, no. But mojob.py doesn't grok platform-specific unit tests, as the
previous CL unintentionally revealed :-)

We've already had a regression due to these tests not being run. I'm open to
suggestions. This course of action (empty test binary on unsupported platform)
is what Trung suggested when I asked him for advice on Friday.