Handle LibraryLoader error from the renderer process.

Handle LibraryLoader error from the renderer process.

If we fail to load the library in the renderer process, just abort early.
This prevents an extra user-visible dialog that Chrome has crashed. Note
that we currently assume the browser process hits the same error and will
prompt the user. If this only happens for the renderer process we'd end up
with a hung tab, but it's not clear how that would happen.

As part of this, convert the RESULT_CODES enum to be generated from a template so that we can re-use the same values for C++ and Java

BUG=180054
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=186610

[Ted C, 2013-03-05 18:41:04]

lgtm

[Yaron, 2013-03-05 21:30:19]

joth: aw
jam: content/public changes (for making ResultCodes available to Java)

Updated based on Michael Bai's suggestions downstream to convert the
UnsatisfiedLinkError to our existing checked exception

https://codereview.chromium.org/12464004/diff/3001/content/public/common/resu...
File content/public/common/result_codes_list.h (right):

https://codereview.chromium.org/12464004/diff/3001/content/public/common/resu...
content/public/common/result_codes_list.h:30: // Failed to register JNI methods.
(Android)
Can't use #if defined(OS_ANDROID) guards because they won't be defined when
generating the java template.

[michaelbai, 2013-03-05 21:44:59]

https://codereview.chromium.org/12464004/diff/3001/content/public/common/resu...
File content/public/common/result_codes_list.h (right):

https://codereview.chromium.org/12464004/diff/3001/content/public/common/resu...
content/public/common/result_codes_list.h:30: // Failed to register JNI methods.
(Android)
Did you mean we didn't defined OS_ANDROID when generate java file? Maybe we
should fix this issue instead of removing the OS_ANDROID guards?

It weird that not defining OS_ANDROID when generate java code since Java is only
used in android.


On 2013/03/05 21:30:20, Yaron wrote:
> Can't use #if defined(OS_ANDROID) guards because they won't be defined when
> generating the java template.

[joth, 2013-03-05 21:49:53]

aw/ lgtm with small comment...

https://codereview.chromium.org/12464004/diff/3001/android_webview/java/src/o...
File android_webview/java/src/org/chromium/android_webview/AwBrowserProcess.java
(right):

https://codereview.chromium.org/12464004/diff/3001/android_webview/java/src/o...
android_webview/java/src/org/chromium/android_webview/AwBrowserProcess.java:40:
throw new RuntimeException("Cannot initialize WebView", e);
nit: initialize -> load

[Yaron, 2013-03-05 22:05:04]

https://codereview.chromium.org/12464004/diff/3001/android_webview/java/src/o...
File android_webview/java/src/org/chromium/android_webview/AwBrowserProcess.java
(right):

https://codereview.chromium.org/12464004/diff/3001/android_webview/java/src/o...
android_webview/java/src/org/chromium/android_webview/AwBrowserProcess.java:40:
throw new RuntimeException("Cannot initialize WebView", e);
On 2013/03/05 21:49:53, joth wrote:
> nit: initialize -> load

Done.

https://codereview.chromium.org/12464004/diff/3001/content/public/common/resu...
File content/public/common/result_codes_list.h (right):

https://codereview.chromium.org/12464004/diff/3001/content/public/common/resu...
content/public/common/result_codes_list.h:30: // Failed to register JNI methods.
(Android)
On 2013/03/05 21:44:59, michaelbai wrote:
> Did you mean we didn't defined OS_ANDROID when generate java file? Maybe we
> should fix this issue instead of removing the OS_ANDROID guards?
> 
> It weird that not defining OS_ANDROID when generate java code since Java is
only
> used in android.
> 
> 
> On 2013/03/05 21:30:20, Yaron wrote:
> > Can't use #if defined(OS_ANDROID) guards because they won't be defined when
> > generating the java template.
> 

I figured it's questionable utility since it's not like we'll re-use 4 for
another port but I guess it's a valid concern so that other ports don't have
that constant defined to erroneously be used. Updated gyp definition so this
works now.

[michaelbai, 2013-03-05 22:07:48]

lgtm

https://codereview.chromium.org/12464004/diff/3001/content/public/common/resu...
File content/public/common/result_codes_list.h (right):

https://codereview.chromium.org/12464004/diff/3001/content/public/common/resu...
content/public/common/result_codes_list.h:30: // Failed to register JNI methods.
(Android)
Thanks

On 2013/03/05 22:05:04, Yaron wrote:
> On 2013/03/05 21:44:59, michaelbai wrote:
> > Did you mean we didn't defined OS_ANDROID when generate java file? Maybe we
> > should fix this issue instead of removing the OS_ANDROID guards?
> > 
> > It weird that not defining OS_ANDROID when generate java code since Java is
> only
> > used in android.
> > 
> > 
> > On 2013/03/05 21:30:20, Yaron wrote:
> > > Can't use #if defined(OS_ANDROID) guards because they won't be defined
when
> > > generating the java template.
> > 
> 
> I figured it's questionable utility since it's not like we'll re-use 4 for
> another port but I guess it's a valid concern so that other ports don't have
> that constant defined to erroneously be used. Updated gyp definition so this
> works now.

[michaelbai, 2013-03-05 22:16:44]

Just realize another issue, you need put result_codes_list.h in template_deps,
otherwise the java file wouldn't be regenerated if header file changed. refer to

https://codereview.chromium.org/11929032/

[michaelbai, 2013-03-05 22:21:05]

On 2013/03/05 22:16:44, michaelbai wrote:
> Just realize another issue, you need put result_codes_list.h in template_deps,
> otherwise the java file wouldn't be regenerated if header file changed. refer
to
> 
> https://codereview.chromium.org/11929032/


Sorry, you already added it, I were looking another file.

[jam, 2013-03-06 17:23:33]

ugh, this doesn't look nice in the content api...

[Yaron, 2013-03-06 18:09:29]

On 2013/03/06 17:23:33, jam wrote:
> ugh, this doesn't look nice in the content api...

It's not the first case. We already have it in
content/public/common/page_transition_types.h

Note that the point of this is to create enums for both C++ and Java that are
consistent since they're auto-generated. Otherwise, I'd have to use a magic
constant in Java and say "keep in sync with native". The reality is that you
have Java consumers of the content API and I think they should speak the same
language as native consumers.

[jam, 2013-03-06 23:54:23]

On 2013/03/06 18:09:29, Yaron wrote:
> On 2013/03/06 17:23:33, jam wrote:
> > ugh, this doesn't look nice in the content api...
> 
> It's not the first case. We already have it in
> content/public/common/page_transition_types.h
> 
> Note that the point of this is to create enums for both C++ and Java that are
> consistent since they're auto-generated. Otherwise, I'd have to use a magic
> constant in Java and say "keep in sync with native". The reality is that you
> have Java consumers of the content API and I think they should speak the same
> language as native consumers.

I see. I forgot that this was already in the content api. ok, lgtm. although i
really wish there was a better way

[commit-bot: I haz the power, 2013-03-07 00:47:12]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/yfriedman@chromium.org/12464004/14002

[commit-bot: I haz the power, 2013-03-07 00:47:14]

Failed to apply patch for build/android/java_cpp_template.gypi:
While running patch -p1 --forward --force --no-backup-if-mismatch;
  patching file build/android/java_cpp_template.gypi
  Hunk #1 FAILED at 58.
  1 out of 1 hunk FAILED -- saving rejects to file
build/android/java_cpp_template.gypi.rej

Patch:       build/android/java_cpp_template.gypi
Index: build/android/java_cpp_template.gypi
diff --git a/build/android/java_cpp_template.gypi
b/build/android/java_cpp_template.gypi
index
89058b0bc9e20b761df442de864dbe453ec4488e..1417d1d80de2de57aaf082dbe38b83bcdc2b1dfc
100644
--- a/build/android/java_cpp_template.gypi
+++ b/build/android/java_cpp_template.gypi
@@ -58,6 +58,7 @@
       ],
       'action': [
         'cpp',                 # invoke host pre-processor.
+        '-D', 'ANDROID',       # Specify ANDROID define for pre-processor.
         '-x', 'c-header',      # treat sources as C header files
         '-P',                  # disable line markers, i.e. '#line 309'
         '-I', '<(DEPTH)',      # Add project top-level to include path

[commit-bot: I haz the power, 2013-03-07 01:10:23]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/yfriedman@chromium.org/12464004/21001

[commit-bot: I haz the power, 2013-03-07 04:12:19]

Change committed as 186610