Up the warnings in ipc (take 2)...

Up the warnings in ipc (take 2)
- add chromium_code:1 to the GYP file
- Fix some unittest compares of literal 0 to apis that return size_t
- initializer order match declared order
- type_id is a uint32, so fix up comparison warnings by using the right type in the test code.
- duplicate a type cast used in the ipc headers into the ipc impl to make windows happy.
- msvc warns about getenv, avoid it.

BUG=none
TEST=everything still builds/works

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=53468

[TVL, 2010-07-02 16:28:20]

seems to pass a clobber win try bot now

[Mark Mentovai, 2010-07-02 16:38:17]

LGTM

[TVL, 2010-07-23 14:14:32]

Updated to today's trunk, has now pass windows and linux try bots.

[Mark Mentovai, 2010-07-23 15:07:28]

LGO. I have a question, though.

http://codereview.chromium.org/2821028/diff/10001/11004
File ipc/ipc_logging.cc (right):

http://codereview.chromium.org/2821028/diff/10001/11004#newcode58
ipc/ipc_logging.cc:58: bool logging_env_var_set = (requiredSize != 0);
I don’t know if requiredSize will be 0 (or something else) if CHROME_IPC_LOGGING
is set but empty. If it’s set but empty, we should still have
logging_env_var_set true, as in the POSIX case.

http://codereview.chromium.org/2821028/diff/10001/11004#newcode60
ipc/ipc_logging.cc:60: bool logging_env_var_set = (getenv("CHROME_IPC_LOGGING")
!= NULL);
You could have a const char* just outside the #if to hold the CHROME_IPC_LOGGING
name so that it’s guaranteed the same for both branches of the #if.

[TVL, 2010-07-23 15:26:55]

http://codereview.chromium.org/2821028/diff/10001/11004
File ipc/ipc_logging.cc (right):

http://codereview.chromium.org/2821028/diff/10001/11004#newcode58
ipc/ipc_logging.cc:58: bool logging_env_var_set = (requiredSize != 0);
On 2010/07/23 15:07:28, Mark Mentovai wrote:
> I don’t know if requiredSize will be 0 (or something else) if
CHROME_IPC_LOGGING
> is set but empty. If it’s set but empty, we should still have
> logging_env_var_set true, as in the POSIX case.

From what msdn says, if you fetch the value, it will come back with a \0 on the
end. So setting it to no value, will resulting in it needing 1 byte to hold the
string terminator for that empty value.