Bug 137660 – [EFL] Add message APIs to communicate between ewk_context and extensions

• Component: WebKit EFL
• Status: RESOLVED FIXED

[Ryuan Choi]

SSIA

[Ryuan Choi]

Created attachment 239735 [details]
Patch

[Ryuan Choi]

Gyuyoung, jinwoo,

Could you take a look at this?

[Ryuan Choi]

Created attachment 241162 [details]
rebased

[Gyuyoung Kim]

Comment on attachment 241162 [details]
rebased

View in context: https://bugs.webkit.org/attachment.cgi?id=241162&action=review

Could you explain what do you want to do in this patch ?

> Source/WebKit2/UIProcess/API/efl/ewk_context.cpp:528
> +        eina_value_get(body, &value);

Why does messageBody need to get eina_value_get() ?

[Ryuan Choi]

Comment on attachment 241162 [details]
rebased

View in context: https://bugs.webkit.org/attachment.cgi?id=241162&action=review

This is the part of ewk_extension which wraps InjectedBundle as ewk API.
This patch provides some APIs of ewk_extension to communicate between ewk_context(UIProcess) and ewk_extension(WebProcess).

Without this, we don't have a way to do it without exposing InjectedBundle and WK C-APIs like tizen did.

>> Source/WebKit2/UIProcess/API/efl/ewk_context.cpp:528
>> +        eina_value_get(body, &value);
> 
> Why does messageBody need to get eina_value_get() ?

It's to support various types in the future although this patch just implements string type.

For example, we can support double or integer type or any structures which EFL supports to post from UIProcess to WebProcess.

[Gyuyoung Kim]

Comment on attachment 241162 [details]
rebased

View in context: https://bugs.webkit.org/attachment.cgi?id=241162&action=review

> Source/WebKit2/UIProcess/API/efl/ewk_context.cpp:411
> +    if (messageBody && WKStringGetTypeID() == WKGetTypeID(messageBody)) {

Can messageBody be null ? Isn't WKTypeRef a reference, not a pointer ?

> Source/WebKit2/UIProcess/API/efl/ewk_context.cpp:526
> +    if (type == EINA_VALUE_TYPE_STRINGSHARE || type == EINA_VALUE_TYPE_STRING) {

Looks like early return is better.

if (type != EINA_VALUE_TYPE_STRINGSHARE && type != EINA_VALUE_TYPE_STRING)
    return false;

const char* value;
eina_value_get(body, &value);
messageBody = adoptWK(WKStringCreateWithUTF8CString(value));

> Source/WebKit2/WebProcess/InjectedBundle/API/efl/ewk_extension.cpp:89
> +    if (messageBody && WKStringGetTypeID() == WKGetTypeID(messageBody)) {

ditto.

> Source/WebKit2/WebProcess/InjectedBundle/API/efl/ewk_extension.cpp:137
> +    if (type == EINA_VALUE_TYPE_STRINGSHARE || type == EINA_VALUE_TYPE_STRING) {

ditto.

[Ryuan Choi]

Created attachment 247213 [details]
Patch

[Ryuan Choi]

(In reply to comment #6)
> Comment on attachment 241162 [details]
> rebased
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=241162&action=review
> 
> > Source/WebKit2/UIProcess/API/efl/ewk_context.cpp:411
> > +    if (messageBody && WKStringGetTypeID() == WKGetTypeID(messageBody)) {
> 
> Can messageBody be null ? Isn't WKTypeRef a reference, not a pointer ?
> 

Yes, WKTypeRef is defined like below in Source/WebKit2/Shared/API/c/WKBase.h
typedef const void* WKTypeRef;

Although we don't pass the empty message logically, I think that this check routine is reasonable.

> > Source/WebKit2/UIProcess/API/efl/ewk_context.cpp:526
> > +    if (type == EINA_VALUE_TYPE_STRINGSHARE || type == EINA_VALUE_TYPE_STRING) {
> 
> Looks like early return is better.
> 
Sure, I fixed.

> if (type != EINA_VALUE_TYPE_STRINGSHARE && type != EINA_VALUE_TYPE_STRING)
>     return false;
> 
> const char* value;
> eina_value_get(body, &value);
> messageBody = adoptWK(WKStringCreateWithUTF8CString(value));
> 
> > Source/WebKit2/WebProcess/InjectedBundle/API/efl/ewk_extension.cpp:89
> > +    if (messageBody && WKStringGetTypeID() == WKGetTypeID(messageBody)) {
> 
> ditto.
> 
> > Source/WebKit2/WebProcess/InjectedBundle/API/efl/ewk_extension.cpp:137
> > +    if (type == EINA_VALUE_TYPE_STRINGSHARE || type == EINA_VALUE_TYPE_STRING) {
> 
> ditto.

[WebKit Commit Bot]

Comment on attachment 247213 [details]
Patch

Clearing flags on attachment: 247213

Committed r180553: <http://trac.webkit.org/changeset/180553>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.