Chrome-side change to support stdout/err to postmessage debug hookup.

Chrome-side change to support stdout/err to postmessage debug hookup.

This CL adds virtual function in reverse interface needed to support
Native Client side debuggging only support for sending standard output
and standard error output (write syscall buffer contents) through the
reverse channel to the renderer to be transformed into PostMessage
data.  We do not use message channels because PPAPI currently does not
support message channels; instead messages are prefixed by a fixed
string which a debug-use-only event listener should match (and invoke
stopPropagation on).

BUG= http://code.google.com/p/nativeclient/issues/detail?id=2474
TEST= nacl-side CL, see http://codereview.chromium.org/8825007


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=113723

[sehr (please use chromium), 2011-12-07 00:32:48]

lgtm

[commit-bot: I haz the power, 2011-12-07 00:33:07]

No comments yet.

[bsy, 2011-12-07 07:43:02]

PTAL.  win trybot is now happy.

[sehr (please use chromium), 2011-12-07 18:45:46]

On 2011/12/07 07:43:02, bsy wrote:
> PTAL.  win trybot is now happy.

LGTM.

[commit-bot: I haz the power, 2011-12-07 18:47:25]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bsy@google.com/8826005/7001

[commit-bot: I haz the power, 2011-12-07 18:47:33]

Presubmit check for 8826005-7001 failed and returned exit status 1.

Running presubmit commit checks ...

** Presubmit ERRORS **
Missing LGTM from an OWNER for: ppapi/cpp/instance.h

Presubmit checks took 1.2s to calculate.

[dmichael (off chromium), 2011-12-07 19:02:13]

http://codereview.chromium.org/8826005/diff/3001/ppapi/native_client/src/trus...
File ppapi/native_client/src/trusted/plugin/service_runtime.cc (right):

http://codereview.chromium.org/8826005/diff/3001/ppapi/native_client/src/trus...
ppapi/native_client/src/trusted/plugin/service_runtime.cc:57: #endif
Wouldn't it have worked to just move this above the includes?

I'd like to avoid having this kind of thing in our public API headers if we can.
I think a small style violation (having stuff before the includes) is worth it
in this case.

[bsy, 2011-12-07 21:18:32]

http://codereview.chromium.org/8826005/diff/3001/ppapi/native_client/src/trus...
File ppapi/native_client/src/trusted/plugin/service_runtime.cc (right):

http://codereview.chromium.org/8826005/diff/3001/ppapi/native_client/src/trus...
ppapi/native_client/src/trusted/plugin/service_runtime.cc:57: #endif
you were looking at an old change set.  this change is now in instance.h.

because of our policy wrt sorting headers and including-what-you-use rather than
requiring users of headers to know header dependencies, do the partial order
sort, and include headers themselves, it was impossible to undef PostMessage
here or even just before the include of instance_private.h from plugin.h.

[bsy, 2011-12-08 06:59:58]

moved undef to service_runtime.cc so that the public header doesn't have to have
anything to indicate that a name collision is possible/likely.  please take
another look.

[dmichael (off chromium), 2011-12-08 17:14:08]

LGTM

Thanks, and sorry for the trouble. For the record, my goal is not to hide the
problem. Putting an undef in instance.h is undesirable for a couple of reasons:
1) All our end-developers use the NaCl toolchain, so the undef in instance.h
would at best be useless and confusing.
2) Doing this kind of thing in a header is notoriously dangerous. If some other
piece of code tries to use instance.h and also windows.h in some other order or
combination, it might break in other subtle ways that would be more difficult to
fix.

(FWIW Another approach would be to see whether the header that's using windows.h
could be refactored to do a body include. That may be tough.)

[commit-bot: I haz the power, 2011-12-08 23:05:12]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bsy@google.com/8826005/18001

[commit-bot: I haz the power, 2011-12-09 01:30:05]

Change committed as 113723