Add a webkit_glue API for lanching NaCl's sel_ldr. NaCl will call this API v...

Add a webkit_glue API for lanching NaCl's sel_ldr.  NaCl will call this API via
Pepper.  This API is based on an existing piece of code used by NaCl's NPAPI
plugin, which will be removed once we get the Pepper plugin off the ground.

[abarth-chromium, 2010-12-16 23:23:06]

Noel, would you be willing to review this patch from a NaCl point of view?  We
can get someone else to review it from a Chrome point of view if need be.

[abarth-chromium, 2010-12-16 23:26:16]

Adding Brett explicitly in case he wants to see all the pieces of this puzzle.

[abarth-chromium, 2010-12-16 23:37:20]

Looks like we have a problem getting nacl::Handle through PPAPI on Windows.  On
windows, that is just a typedef for HANDLE.

[noelallen_use_chromium, 2010-12-17 19:52:22]

LGTM, however you may want to add that '#ifdef' if compiling chrome without NaCl
is desirable.

http://codereview.chromium.org/5933004/diff/9001/chrome/renderer/render_proce...
File chrome/renderer/render_process_impl.cc (right):

http://codereview.chromium.org/5933004/diff/9001/chrome/renderer/render_proce...
chrome/renderer/render_process_impl.cc:60: int* nacl_process_id) {
Ugly that this is an int, however tracing through the code we see that
sel_ldr_launcher_chrome.cc which actually uses this through func* indirection
passes in an actual int ptr.  So this is correct.

http://codereview.chromium.org/5933004/diff/9001/chrome/renderer/renderer_glu...
File chrome/renderer/renderer_glue.cc (right):

http://codereview.chromium.org/5933004/diff/9001/chrome/renderer/renderer_glu...
chrome/renderer/renderer_glue.cc:31: #include
"native_client/src/shared/imc/nacl_imc.h"
#ifdef this as well?  So that someone could compile without nacl.

[brettw, 2010-12-17 20:42:25]

I wonder, can this be implemented via the PepperPluginDelegate rather than the
renderer glue stuff? I think this is preferrable since it groups all the Pepper
stuff and I think the "bunch of global functions" is kind of ugly.

[abarth-chromium, 2010-12-17 20:47:20]

On 2010/12/17 20:42:25, brettw wrote:
> I wonder, can this be implemented via the PepperPluginDelegate rather than the
> renderer glue stuff? I think this is preferrable since it groups all the
Pepper
> stuff and I think the "bunch of global functions" is kind of ugly.

Sure, we could do that if you like.  The trade-off is how much you think this
stuff is Pepper-related as opposed to more generic glue.

[brettw, 2010-12-17 20:53:59]

On Fri, Dec 17, 2010 at 12:47 PM,  <abarth@chromium.org> wrote:
> On 2010/12/17 20:42:25, brettw wrote:
>>
>> I wonder, can this be implemented via the PepperPluginDelegate rather than
>> the
>> renderer glue stuff? I think this is preferrable since it groups all the
>
> Pepper
>>
>> stuff and I think the "bunch of global functions" is kind of ugly.
>
> Sure, we could do that if you like.  The trade-off is how much you think
> this
> stuff is Pepper-related as opposed to more generic glue.

We've done a bunch of Flash specific goop in there, so I think all the
"special Pepper stuff" should just go there.

Brett

[abarth-chromium, 2010-12-17 20:55:09]

On Fri, Dec 17, 2010 at 12:53 PM, Brett Wilson <brettw@chromium.org> wrote:
> On Fri, Dec 17, 2010 at 12:47 PM,  <abarth@chromium.org> wrote:
>> On 2010/12/17 20:42:25, brettw wrote:
>>>
>>> I wonder, can this be implemented via the PepperPluginDelegate rather than
>>> the
>>> renderer glue stuff? I think this is preferrable since it groups all the
>>
>> Pepper
>>>
>>> stuff and I think the "bunch of global functions" is kind of ugly.
>>
>> Sure, we could do that if you like.  The trade-off is how much you think
>> this
>> stuff is Pepper-related as opposed to more generic glue.
>
> We've done a bunch of Flash specific goop in there, so I think all the
> "special Pepper stuff" should just go there.

Ok.  Will do.

Adam