[NaCl SDK] Remove invalid assert from nacl_io.

[NaCl SDK] Remove invalid assert from nacl_io.

Also, add NDEBUG to CFLAGS by default so that asserts not
compiled into release builds, and fix resulting unused
variable warnings by making KernelProxy::Init do more
error checking and return a result.

R=binji@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=221470

[Sam Clegg, 2013-08-24 01:23:54]

Had to do a little more to this change to handle the used
variable in kernel_proxy.cc.

[binji, 2013-08-24 01:35:57]

lgtm

https://codereview.chromium.org/22842011/diff/12001/native_client_sdk/src/lib...
File native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc (right):

https://codereview.chromium.org/22842011/diff/12001/native_client_sdk/src/lib...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:102: if (result != 0)
maybe use assert(false) inside the if instead?

[Sam Clegg, 2013-09-04 19:10:34]

fix input_event

[Sam Clegg, 2013-09-04 19:10:58]

Fixed a couple of examples.

PTAL.

[binji, 2013-09-04 20:21:47]

slgtm

You'll need to try against HEAD, I think. The lkgr is broken right now

https://codereview.chromium.org/22842011/diff/43001/native_client_sdk/src/exa...
File native_client_sdk/src/examples/api/input_event/shared_queue.h (right):

https://codereview.chromium.org/22842011/diff/43001/native_client_sdk/src/exa...
native_client_sdk/src/examples/api/input_event/shared_queue.h:12: #ifdef
__GNUC__
not sure I like adding UNUSED to this header. Does it fix the warning if you do
this instead?

...
assert(result == 0);
(void)result;

or maybe disable the warning (via #pragma) below?
or move the #ifdef below?

https://codereview.chromium.org/22842011/diff/43001/native_client_sdk/src/lib...
File native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc (right):

https://codereview.chromium.org/22842011/diff/43001/native_client_sdk/src/lib...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:118: rtn = fd;
rtn = errno?

[Sam Clegg, 2013-09-04 20:54:43]

https://codereview.chromium.org/22842011/diff/43001/native_client_sdk/src/exa...
File native_client_sdk/src/examples/api/input_event/shared_queue.h (right):

https://codereview.chromium.org/22842011/diff/43001/native_client_sdk/src/exa...
native_client_sdk/src/examples/api/input_event/shared_queue.h:12: #ifdef
__GNUC__
On 2013/09/04 20:21:48, binji wrote:
> not sure I like adding UNUSED to this header. Does it fix the warning if you
do
> this instead?
> 
> ...
> assert(result == 0);
> (void)result;
> 
> or maybe disable the warning (via #pragma) below?
> or move the #ifdef below?


I prefer the attribute to the #pragma.

(void) doesn't fix the issue in my experience. The compilers are
getting too clever.  Even if it did work is less explicit and 
more fragile.

Moved the macro down.

https://codereview.chromium.org/22842011/diff/43001/native_client_sdk/src/lib...
File native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc (right):

https://codereview.chromium.org/22842011/diff/43001/native_client_sdk/src/lib...
native_client_sdk/src/libraries/nacl_io/kernel_proxy.cc:118: rtn = fd;
On 2013/09/04 20:21:48, binji wrote:
> rtn = errno?

Done.

[binji, 2013-09-04 22:22:31]

https://codereview.chromium.org/22842011/diff/25001/native_client_sdk/src/exa...
File native_client_sdk/src/examples/api/input_event/shared_queue.h (right):

https://codereview.chromium.org/22842011/diff/25001/native_client_sdk/src/exa...
native_client_sdk/src/examples/api/input_event/shared_queue.h:57: #define UNUSED
__attribute__ ((unused))
sorry, I meant use __attribute__ without defining a new macro. The thing I'm
concerned about is polluting the macro namespace by including a header.

[Sam Clegg, 2013-09-05 18:33:03]

Committed patchset #9 manually as r221470 (presubmit successful).