Bug 135606 – [EFL] Prevent the client from creating ewk_view when EWebkit is not initialized

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit EFL

[Grzegorz Czajkowski]

Similarly to EFL modules (eina, evas etc.), application using EWebKit has to initialize it using ewk_init(). Do not allow the client to create ewk_view if ewk_init has not been called.

[Grzegorz Czajkowski]

Created attachment 236028 [details]
Patch

[Gyuyoung Kim]

Comment on attachment 236028 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=236028&action=review

Thank you for nice fix !

> Source/WebKit2/UIProcess/efl/EwkMain.cpp:3
> +    Copyright (C) 2009-2011 Samsung Electronics

Please update 2011 -> 2014.

> Source/WebKit2/UIProcess/efl/EwkMain.cpp:73
> +        EINA_LOG_CRIT("could not register log domain 'ewebkit2'");

I hope to use same macro as below. CRITICAL()

> Source/WebKit2/UIProcess/efl/EwkMain.cpp:80
> +        eina_log_domain_unregister(m_ewkLogDomainId);

How about adding a function to process shutdown ? For example,

if (!evas_init()) {
    CRITICAL("could not init evas.");
    shutdownEFLLibraries(EVAS_INIT);
}

if (!ecore_init()) {
    CRITICAL("could not init ecore.");
    shutdownEFLLibraries(EVAS_INIT);
}

EwkMain::shutdownEFLLibraries(int failed) {
    switch(failed) {
    case ecore_init:
        break;
    case ecore_evas_init:
        break;
    case ecore_imf_init:
        ecore_evas_shutdown();
        ecore_shutdown();
        evas_shutdown();
        break;
    case efreet_init:
        break;
    }

    eina_log_domain_unregister(m_ewkLogDomainId);
    m_ewkLogDomainId = -1;
    eina_shutdown();
    return 0;
}

[Ryuan Choi]

Comment on attachment 236028 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=236028&action=review

IMHO, EwkMain in ewk_main_private.h looks fine (without WK).

> Source/WebKit2/UIProcess/API/C/efl/WKMain.cpp:33
> +int WKMainInitialize()

Why WKxxx ?

It looks not quite related to WK APIs to me.

> Source/WebKit2/UIProcess/efl/EwkMain.cpp:34
> +#include <glib-object.h>
> +#include <glib.h>

Should we still include this?

> Source/WebKit2/UIProcess/efl/EwkMain.cpp:64
> +    if (isInitialized())
> +        return ++m_ewkInitCount;

Nit,
Just for me,
if (m_ewkInitCount) looks readable.

[Grzegorz Czajkowski]

Created attachment 236094 [details]
Apply Gyuyoung and Ryuan comments

[Grzegorz Czajkowski]

Comment on attachment 236028 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=236028&action=review

Thank you guys for the review!

>> Source/WebKit2/UIProcess/API/C/efl/WKMain.cpp:33
>> +int WKMainInitialize()
> 
> Why WKxxx ?
> 
> It looks not quite related to WK APIs to me.

Yeahh, that was overkill a bit. What I wanted to achieve was to use WK C API (WKMainIsInitialized) in ewk_view.cpp. It seems that the style in ewk is completely mixed.

>> Source/WebKit2/UIProcess/efl/EwkMain.cpp:3
>> +    Copyright (C) 2009-2011 Samsung Electronics
> 
> Please update 2011 -> 2014.

Done.

>> Source/WebKit2/UIProcess/efl/EwkMain.cpp:34
>> +#include <glib.h>
> 
> Should we still include this?

Good catch! Indeed they are not needed any longer.

>> Source/WebKit2/UIProcess/efl/EwkMain.cpp:64
>> +        return ++m_ewkInitCount;
> 
> Nit,
> Just for me,
> if (m_ewkInitCount) looks readable.

Restored.

>> Source/WebKit2/UIProcess/efl/EwkMain.cpp:73
>> +        EINA_LOG_CRIT("could not register log domain 'ewebkit2'");
> 
> I hope to use same macro as below. CRITICAL()

Unfortunately we ca not use CRITICAL() macro here as the registration of new log domain (which CRITICAL uses) has failed. Similarly to !eina_init() - fixed in patch #2.

>> Source/WebKit2/UIProcess/efl/EwkMain.cpp:80
>> +        eina_log_domain_unregister(m_ewkLogDomainId);
> 
> How about adding a function to process shutdown ? For example,
> 
> if (!evas_init()) {
>     CRITICAL("could not init evas.");
>     shutdownEFLLibraries(EVAS_INIT);
> }
> 
> if (!ecore_init()) {
>     CRITICAL("could not init ecore.");
>     shutdownEFLLibraries(EVAS_INIT);
> }
> 
> EwkMain::shutdownEFLLibraries(int failed) {
>     switch(failed) {
>     case ecore_init:
>         break;
>     case ecore_evas_init:
>         break;
>     case ecore_imf_init:
>         ecore_evas_shutdown();
>         ecore_shutdown();
>         evas_shutdown();
>         break;
>     case efreet_init:
>         break;
>     }
> 
>     eina_log_domain_unregister(m_ewkLogDomainId);
>     m_ewkLogDomainId = -1;
>     eina_shutdown();
>     return 0;
> }

That's is cool! Added. Thanks!

[Ryuan Choi]

Comment on attachment 236094 [details]
Apply Gyuyoung and Ryuan comments

View in context: https://bugs.webkit.org/attachment.cgi?id=236094&action=review

> Source/WebKit2/UIProcess/API/efl/ewk_main_private.h:31
> +typedef unsigned EFLLibraryInitFailure;

How about using enum class or exposing enum ?

> Source/WebKit2/UIProcess/API/efl/ewk_main_private.h:48
> +    int m_ewkInitCount;
> +    int m_ewkLogDomainId;

ewk looks meaningless.

> Source/WebKit2/UIProcess/API/efl/ewk_private.h:29
> +#include "ewk_main_private.h"

Hmm. should we keep ewk_private.h?

[Gyuyoung Kim]

Comment on attachment 236094 [details]
Apply Gyuyoung and Ryuan comments

View in context: https://bugs.webkit.org/attachment.cgi?id=236094&action=review

> Source/WebKit2/UIProcess/API/efl/ewk_main.cpp:189
> +int ewk_init(void)

IIRC, we decided that void parameter is only used for webkit-efl's public header.

> Source/WebKit2/UIProcess/API/efl/ewk_main.cpp:194
> +int ewk_shutdown(void)

ditto.

[Grzegorz Czajkowski]

Comment on attachment 236094 [details]
Apply Gyuyoung and Ryuan comments

View in context: https://bugs.webkit.org/attachment.cgi?id=236094&action=review

Thanks for the review. I will take it into consideration in patch #3.

>> Source/WebKit2/UIProcess/API/efl/ewk_private.h:29
>> +#include "ewk_main_private.h"
> 
> Hmm. should we keep ewk_private.h?

It seems that ewk_private.h is responsible for defining debug macros. How about renaming it to EwkDebug,h and moving out of API/ directory?

[Ryuan Choi]

(In reply to comment #8)
> (From update of attachment 236094 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=236094&action=review
> 
> Thanks for the review. I will take it into consideration in patch #3.
> 
> >> Source/WebKit2/UIProcess/API/efl/ewk_private.h:29
> >> +#include "ewk_main_private.h"
> > 
> > Hmm. should we keep ewk_private.h?
> 
> It seems that ewk_private.h is responsible for defining debug macros. How about renaming it to EwkDebug,h and moving out of API/ directory?

+1

[Grzegorz Czajkowski]

Created attachment 236364 [details]
Patch

[Grzegorz Czajkowski]

(In reply to comment #9)
> (In reply to comment #8)
> > (From update of attachment 236094 [details] [details])
> > View in context: https://bugs.webkit.org/attachment.cgi?id=236094&action=review
> > 
> > Thanks for the review. I will take it into consideration in patch #3.
> > 
> > >> Source/WebKit2/UIProcess/API/efl/ewk_private.h:29
> > >> +#include "ewk_main_private.h"
> > > 
> > > Hmm. should we keep ewk_private.h?
> > 
> > It seems that ewk_private.h is responsible for defining debug macros. How about renaming it to EwkDebug,h and moving out of API/ directory?
> 
> +1

I'll do it separately at bug 135797.

[Gyuyoung Kim]

Comment on attachment 236364 [details]
Patch

LGTM

[Grzegorz Czajkowski]

Committed r172430: <http://trac.webkit.org/changeset/172430>