Add ipc_channel_nacl to the untrusted build and fix all compile/link errors.

Add ipc_channel_nacl to the untrusted build and fix all compile/link errors.
ppapi_proxy_untrusted.gyp should now build with a single link error for main.
BUG=116317
TEST=builds with a single link error for main().
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=133519

[dmichael (off chromium), 2012-04-23 14:51:11]

http://codereview.chromium.org/10167024/diff/1/ipc/ipc_channel.cc
File ipc/ipc_channel.cc (left):

http://codereview.chromium.org/10167024/diff/1/ipc/ipc_channel.cc#oldcode34
ipc/ipc_channel.cc:34: base::GetCurrentProcId(),
Would it be simpler to give base::GetCurrentProcId a stub implementation?

http://codereview.chromium.org/10167024/diff/1/ipc/ipc_channel_nacl.h
File ipc/ipc_channel_nacl.h (left):

http://codereview.chromium.org/10167024/diff/1/ipc/ipc_channel_nacl.h#oldcode28
ipc/ipc_channel_nacl.h:28: int GetClientFileDescriptor() const;
Why non-const? I'd rather leave it as-is if possible.

[bbudge, 2012-04-23 17:46:09]

http://codereview.chromium.org/10167024/diff/1/ipc/ipc_channel.cc
File ipc/ipc_channel.cc (left):

http://codereview.chromium.org/10167024/diff/1/ipc/ipc_channel.cc#oldcode34
ipc/ipc_channel.cc:34: base::GetCurrentProcId(),
base/process_util.cc wouldn't build untrusted unless we commented out almost all
of it. Instead, I think it's better this way, as we can more easily see where
we're substituting '0' in case of problems.
On 2012/04/23 14:51:11, dmichael wrote:
> Would it be simpler to give base::GetCurrentProcId a stub implementation?

http://codereview.chromium.org/10167024/diff/1/ipc/ipc_channel_nacl.h
File ipc/ipc_channel_nacl.h (left):

http://codereview.chromium.org/10167024/diff/1/ipc/ipc_channel_nacl.h#oldcode28
ipc/ipc_channel_nacl.h:28: int GetClientFileDescriptor() const;
Good catch, I removed it erroneously while chasing down compiler errors.
On 2012/04/23 14:51:11, dmichael wrote:
> Why non-const? I'd rather leave it as-is if possible.

[dmichael (off chromium), 2012-04-23 17:46:44]

lgtm

[brettw, 2012-04-23 21:17:35]

lgtm

http://codereview.chromium.org/10167024/diff/10001/ipc/ipc_channel_nacl.cc
File ipc/ipc_channel_nacl.cc (right):

http://codereview.chromium.org/10167024/diff/10001/ipc/ipc_channel_nacl.cc#ne...
ipc/ipc_channel_nacl.cc:68: Channel::ChannelImpl::ReadData(char* buffer,
I think generally we wouldn't indent this second line any.

http://codereview.chromium.org/10167024/diff/10001/ipc/ipc_channel_nacl.cc#ne...
ipc/ipc_channel_nacl.cc:144: base::ProcessId Channel::peer_pid() const { return
0; }
I'd do multiple lines except for some cases in a class definition.