Bug 61850 – [EFL][WK2] Add MiniBrowserEfl.

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit2

[Ryuan Choi]

Patch coming.
It will work when all of WebKit2/EFL land.

[Ryuan Choi]

Created attachment 95572 [details]
Patch

[Gyuyoung Kim]

I think it is better to insert CMakeListEfl.txt into efl directory. MiniBrowser of gtk, qt have build script in its directory.

[Lucas De Marchi]

Comment on attachment 95572 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=95572&action=review

Once WebKit2 is compiled in EFL port, I think this can enter. I agree with Gyuyoung that the cmake file should be in efl directory. But we can do a bit better and split it in a generic CMakeLists.txt file and another efl/CMakeListsEfl.txt. This way other ports that use cmake can profit from this work too.

> Tools/MiniBrowser/CMakeListsEfl.txt:15
> +    ${WEBKIT2_DIR}/UIProcess/API/efl
> +    ${WEBKIT2_DIR}
> +    ${DERIVED_SOURCES_WEBKIT2_DIR}/include

Since we don't support webkit2, are you sure you need this?

> Tools/MiniBrowser/efl/main.c:55
> +static int browserCreate(const char *url, const char *engine)
> +{
> +    MiniBrowser *app = (MiniBrowser*) malloc(sizeof(MiniBrowser));
> +
> +    app->ee = ecore_evas_software_x11_new(NULL, 0, 0, 0, DEFAULT_WIDTH, DEFAULT_HEIGHT);

Could you use ecore_evas_new() here instead?

> Tools/MiniBrowser/efl/main.c:56
> +    ecore_evas_title_set(app->ee, "Samsung MiniBrowser");

I think it's not good to let Samsung here. WebKit would be a better name.

> Tools/MiniBrowser/efl/main.c:71
> +    /* Create webview */
> +    app->browser = ewk_view_add(app->evas, WKContextGetSharedProcessContext(), 0);

It seems you are indeed using webkit2. I'm wondering what's happening since we don't have support to webkit2 yet.

> Tools/MiniBrowser/efl/main.c:97
> +    browserCreate(url, engine);
> +
> +    ecore_event_handler_add(ECORE_EVENT_SIGNAL_EXIT, mainSignalExit, NULL);

Since in this file we are following EFL style, I think it's better not to use camelCase for function names.

[Lucas De Marchi]

(In reply to comment #3)
> (From update of attachment 95572 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=95572&action=review
> 
> Once WebKit2 is compiled in EFL port, I think this can enter. I agree with Gyuyoung that the cmake file should be in efl directory. But we can do a bit better and split it in a generic CMakeLists.txt file and another efl/CMakeListsEfl.txt. This way other ports that use cmake can profit from this work too.
> 
> > Tools/MiniBrowser/CMakeListsEfl.txt:15
> > +    ${WEBKIT2_DIR}/UIProcess/API/efl
> > +    ${WEBKIT2_DIR}
> > +    ${DERIVED_SOURCES_WEBKIT2_DIR}/include
> 
> Since we don't support webkit2, are you sure you need this?

I've just seen you already answered this on your first comment. Sorry for the noise.

[Gyuyoung Kim]

(In reply to comment #3)
> It seems you are indeed using webkit2. I'm wondering what's happening since we don't have support to webkit2 yet.

We start to support WebKit2 for EFL port. We're going to notify this news to webkit-dev mailing list.

[Ryuan Choi]

Created attachment 95591 [details]
Patch

[WebKit Review Bot]

Attachment 95591 [details] did not pass style-queue:

Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'Tools/ChangeLog', u'Tools/MiniBrowser/efl/..." exit_code: 1

Tools/MiniBrowser/efl/main.c:45:  main_signal_exit is incorrectly named. Don't use underscores in your identifier names.  [readability/naming] [4]
Total errors found: 1 in 3 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Ryuan Choi]

(In reply to comment #2)
> I think it is better to insert CMakeListEfl.txt into efl directory. MiniBrowser of gtk, qt have build script in its directory.

It's my mistake. I moved it into efl directory.

(In reply to comment #3)
> (From update of attachment 95572 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=95572&action=review
> 
> Once WebKit2 is compiled in EFL port, I think this can enter. I agree with Gyuyoung that the cmake file should be in efl directory. But we can do a bit better and split it in a generic CMakeLists.txt file and another efl/CMakeListsEfl.txt. This way other ports that use cmake can profit from this work too.
> 
Done.
It's my mistake.

> > Tools/MiniBrowser/CMakeListsEfl.txt:15
> > +    ${WEBKIT2_DIR}/UIProcess/API/efl
> > +    ${WEBKIT2_DIR}
> > +    ${DERIVED_SOURCES_WEBKIT2_DIR}/include
> 
> Since we don't support webkit2, are you sure you need this?
> 
> > Tools/MiniBrowser/efl/main.c:55
> > +static int browserCreate(const char *url, const char *engine)
> > +{
> > +    MiniBrowser *app = (MiniBrowser*) malloc(sizeof(MiniBrowser));
> > +
> > +    app->ee = ecore_evas_software_x11_new(NULL, 0, 0, 0, DEFAULT_WIDTH, DEFAULT_HEIGHT);
> 
> Could you use ecore_evas_new() here instead?
Done.

> 
> > Tools/MiniBrowser/efl/main.c:56
> > +    ecore_evas_title_set(app->ee, "Samsung MiniBrowser");
> 
> I think it's not good to let Samsung here. WebKit would be a better name.
> 
> > Tools/MiniBrowser/efl/main.c:71
> > +    /* Create webview */
> > +    app->browser = ewk_view_add(app->evas, WKContextGetSharedProcessContext(), 0);
> 
> It seems you are indeed using webkit2. I'm wondering what's happening since we don't have support to webkit2 yet.
> 
> > Tools/MiniBrowser/efl/main.c:97
> > +    browserCreate(url, engine);
> > +
> > +    ecore_event_handler_add(ECORE_EVENT_SIGNAL_EXIT, mainSignalExit, NULL);
> 
> Since in this file we are following EFL style, I think it's better not to use camelCase for function names.

I changed it EFL style. but we already use browserCreate. Is it ok?

(In reply to comment #4)
> (In reply to comment #3)
> > (From update of attachment 95572 [details] [details])
> > View in context: https://bugs.webkit.org/attachment.cgi?id=95572&action=review
> > 
> > Once WebKit2 is compiled in EFL port, I think this can enter. I agree with Gyuyoung that the cmake file should be in efl directory. But we can do a bit better and split it in a generic CMakeLists.txt file and another efl/CMakeListsEfl.txt. This way other ports that use cmake can profit from this work too.
> > 
> > > Tools/MiniBrowser/CMakeListsEfl.txt:15
> > > +    ${WEBKIT2_DIR}/UIProcess/API/efl
> > > +    ${WEBKIT2_DIR}
> > > +    ${DERIVED_SOURCES_WEBKIT2_DIR}/include
> > 
> > Since we don't support webkit2, are you sure you need this?
> 
> I've just seen you already answered this on your first comment. Sorry for the noise.

No problem.

[Gyuyoung Kim]

Comment on attachment 95591 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=95591&action=review

Why don't you add WebKit2 prefix to Bug summary ? For example, [EFL/WebKit2] and so on.

> Tools/MiniBrowser/efl/main.c:98
> +    browserCreate(url, engine)

Missing semicolon.

[Ryuan Choi]

Created attachment 95704 [details]
Patch

[WebKit Review Bot]

Attachment 95704 [details] did not pass style-queue:

Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'Tools/ChangeLog', u'Tools/MiniBrowser/efl/..." exit_code: 1

Tools/MiniBrowser/efl/main.c:45:  main_signal_exit is incorrectly named. Don't use underscores in your identifier names.  [readability/naming] [4]
Total errors found: 1 in 3 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Gyuyoung Kim]

Please upload this patch again after landing bug 61903's patch

[Ryuan Choi]

Created attachment 95833 [details]
Patch

[Gyuyoung Kim]

LGTM.

[Lucas De Marchi]

Comment on attachment 95833 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=95833&action=review

> Tools/MiniBrowser/efl/CMakeListsEfl.txt:64
> +ADD_CUSTOM_TARGET(forwarding-headerMiniBrowserSoup
> +    COMMAND ${PERL_EXECUTABLE} ${WEBKIT2_DIR}/Scripts/generate-forwarding-headers.pl ${MiniBrowser_DIR} ${DERIVED_SOURCES_WEBKIT2_DIR}/include soup
> +)

Do you really need this if WTF_USE_CURL is defined?

> Tools/MiniBrowser/efl/CMakeListsEfl.txt:71
> +ADD_DEPENDENCIES(Programs/MiniBrowser forwarding-headerMiniBrowserSoup)

ditto.

[Ryuan Choi]

Created attachment 97075 [details]
Patch

[Ryuan Choi]

(In reply to comment #15)
> (From update of attachment 95833 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=95833&action=review
> 
> > Tools/MiniBrowser/efl/CMakeListsEfl.txt:64
> > +ADD_CUSTOM_TARGET(forwarding-headerMiniBrowserSoup
> > +    COMMAND ${PERL_EXECUTABLE} ${WEBKIT2_DIR}/Scripts/generate-forwarding-headers.pl ${MiniBrowser_DIR} ${DERIVED_SOURCES_WEBKIT2_DIR}/include soup
> > +)
> 
> Do you really need this if WTF_USE_CURL is defined?
> 
> > Tools/MiniBrowser/efl/CMakeListsEfl.txt:71
> > +ADD_DEPENDENCIES(Programs/MiniBrowser forwarding-headerMiniBrowserSoup)
> 
> ditto.

Sorry, I fixed.

[Leandro Pereira]

Comment on attachment 97075 [details]
Patch

Informal r+

[Ryuan Choi]

Created attachment 98267 [details]
Patch_changed_copyright

[Ryuan Choi]

Created attachment 98700 [details]
Patch

[Gyuyoung Kim]

Comment on attachment 98700 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=98700&action=review

> Tools/MiniBrowser/efl/main.c:45
> +    return 1;

Why do you use 1 instead of EINA_TRUE ?

[Ryuan Choi]

Created attachment 98703 [details]
Patch

[Ryuan Choi]

(In reply to comment #21)
> (From update of attachment 98700 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=98700&action=review
> 
> > Tools/MiniBrowser/efl/main.c:45
> > +    return 1;
> 
> Why do you use 1 instead of EINA_TRUE ?

Sorry, It's mistake.
Updated.

[Ryuan Choi]

Comment on attachment 98703 [details]
Patch

This patch is outdated.

I'll upload newer after tested.

[Ryuan Choi]

Created attachment 120781 [details]
Patch

[Raphael Kubo da Costa (:rakuco)]

Comment on attachment 120781 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=120781&action=review

> Tools/MiniBrowser/efl/CMakeLists.txt:17
> +    ${CMAKE_SOURCE_DIR}

Is this one really needed?

> Tools/MiniBrowser/efl/main.c:23
> +#include <Ecore_Getopt.h>
> +#include <Ecore_X.h>

These headers do not seem to be used, and Eina.h is missing.

> Tools/MiniBrowser/efl/main.c:27
> +#define DEFAULT_WIDTH      800
> +#define DEFAULT_HEIGHT     600

static const int would be better than #define.

> Tools/MiniBrowser/efl/main.c:42
> +static Eina_Bool browserCreate(const char *url, const char *engine)

The return type is not being checked; why not make the function void or return a MiniBrowser* so you can free it later?

> Tools/MiniBrowser/efl/main.c:44
> +    MiniBrowser *app = (MiniBrowser*) malloc(sizeof(MiniBrowser));

No need for the cast, and the pointer is being leaked.

> Tools/MiniBrowser/efl/main.c:46
> +    app->ee = ecore_evas_new(engine, 0, 0, DEFAULT_WIDTH, DEFAULT_HEIGHT, NULL);

s/NULL/0/?

> Tools/MiniBrowser/efl/main.c:78
> +    char *engine = NULL;

This is not being used.

> Tools/MiniBrowser/efl/main.c:87
> +        url = (char*) default_url;

Isn't it better to make url const instead of casting?

> Tools/MiniBrowser/efl/main.c:91
> +    ecore_event_handler_add(ECORE_EVENT_SIGNAL_EXIT, main_signal_exit, NULL);

The handler is being leaked.

[Ryuan Choi]

Created attachment 120857 [details]
Patch

[Ryuan Choi]

Created attachment 120860 [details]
Patch

[Ryuan Choi]

Created attachment 120861 [details]
Patch

[Ryuan Choi]

(In reply to comment #29)
> Created an attachment (id=120861) [details]
> Patch

kubo, thank you for your comment.

Fixed all you commented.

[Gyuyoung Kim]

Comment on attachment 120861 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=120861&action=review

> Tools/MiniBrowser/efl/main.c:27
> +static const char *DEFAULT_URL = "http://www.google.com/";

As you may know, this violate *whitespace/declaration* rule. Bug 75424 will remove the rule's exception from EFL port. But, I am not sure if we should implement Mini Browser according to EFL coding style. GTK port implemented it according to GTK coding style.(So, there are style errors gtk MiniBrowser.) But, it seems other ports don't violate webkit coding rule. IMO, why don't we adhere *whitespace/declaration* rule in here ?

[Ryuan Choi]

(In reply to comment #31)
> (From update of attachment 120861 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=120861&action=review
> 
> > Tools/MiniBrowser/efl/main.c:27
> > +static const char *DEFAULT_URL = "http://www.google.com/";
> 
> As you may know, this violate *whitespace/declaration* rule. Bug 75424 will remove the rule's exception from EFL port. But, I am not sure if we should implement Mini Browser according to EFL coding style. GTK port implemented it according to GTK coding style.(So, there are style errors gtk MiniBrowser.) But, it seems other ports don't violate webkit coding rule. IMO, why don't we adhere *whitespace/declaration* rule in here ?

Right, I know that MiniBrowser and EWebLauncher doesn't keep the webkit coding style and it's sad for me.

IMO, We can fix them after getting approval from other webkit/efl guys.

[Gyuyoung Kim]

(In reply to comment #32)
> (In reply to comment #31)
> > (From update of attachment 120861 [details] [details])
> > View in context: https://bugs.webkit.org/attachment.cgi?id=120861&action=review
> > 
> > > Tools/MiniBrowser/efl/main.c:27
> > > +static const char *DEFAULT_URL = "http://www.google.com/";
> > 
> > As you may know, this violate *whitespace/declaration* rule. Bug 75424 will remove the rule's exception from EFL port. But, I am not sure if we should implement Mini Browser according to EFL coding style. GTK port implemented it according to GTK coding style.(So, there are style errors gtk MiniBrowser.) But, it seems other ports don't violate webkit coding rule. IMO, why don't we adhere *whitespace/declaration* rule in here ?
> 
> Right, I know that MiniBrowser and EWebLauncher doesn't keep the webkit coding style and it's sad for me.
> 
> IMO, We can fix them after getting approval from other webkit/efl guys.

My though is changed. EWebLauncher and MiniBrowser are EFL application. So, I think it is better to keep to implement with EFL coding style. In Bug 75424, I keep *whitespace/declaration* exception for EWebLauncher and MiniBrowser.

[Ryuan Choi]

(In reply to comment #33)
> (In reply to comment #32)
> > (In reply to comment #31)
> > IMO, We can fix them after getting approval from other webkit/efl guys.
> 
> My though is changed. EWebLauncher and MiniBrowser are EFL application. So, I think it is better to keep to implement with EFL coding style. In Bug 75424, I keep *whitespace/declaration* exception for EWebLauncher and MiniBrowser.

OK, I love webkit coding style, but I'm not sure which one we should keep it.
I'll follow the rule which the most people like.

Thanks for your changes.

[Raphael Kubo da Costa (:rakuco)]

Comment on attachment 120861 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=120861&action=review

Looks mostly OK even though I'm unable to test it.

>>>> Tools/MiniBrowser/efl/main.c:27
>>>> +static const char *DEFAULT_URL = "http://www.google.com/";
>>> 
>>> As you may know, this violate *whitespace/declaration* rule. Bug 75424 will remove the rule's exception from EFL port. But, I am not sure if we should implement Mini Browser according to EFL coding style. GTK port implemented it according to GTK coding style.(So, there are style errors gtk MiniBrowser.) But, it seems other ports don't violate webkit coding rule. IMO, why don't we adhere *whitespace/declaration* rule in here ?
>> 
>> Right, I know that MiniBrowser and EWebLauncher doesn't keep the webkit coding style and it's sad for me.
>> 
>> IMO, We can fix them after getting approval from other webkit/efl guys.
> 
> My though is changed. EWebLauncher and MiniBrowser are EFL application. So, I think it is better to keep to implement with EFL coding style. In Bug 75424, I keep *whitespace/declaration* exception for EWebLauncher and MiniBrowser.

Another (minor) issue here is that this does not go to the .rodata section in the binary; use static const char foo[] instead.

[Ryuan Choi]

Created attachment 120903 [details]
Patch

[Ryuan Choi]

(In reply to comment #35)
> (From update of attachment 120861 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=120861&action=review
> 
> Looks mostly OK even though I'm unable to test it.
> 
Ah, I shared minimum patch to run webkit2/efl.
you can run minibrowser/efl using it.

> >>>> Tools/MiniBrowser/efl/main.c:27
> >>>> +static const char *DEFAULT_URL = "http://www.google.com/";
> >>> 
> >>> As you may know, this violate *whitespace/declaration* rule. Bug 75424 will remove the rule's exception from EFL port. But, I am not sure if we should implement Mini Browser according to EFL coding style. GTK port implemented it according to GTK coding style.(So, there are style errors gtk MiniBrowser.) But, it seems other ports don't violate webkit coding rule. IMO, why don't we adhere *whitespace/declaration* rule in here ?
> >> 
> >> Right, I know that MiniBrowser and EWebLauncher doesn't keep the webkit coding style and it's sad for me.
> >> 
> >> IMO, We can fix them after getting approval from other webkit/efl guys.
> > 
> > My though is changed. EWebLauncher and MiniBrowser are EFL application. So, I think it is better to keep to implement with EFL coding style. In Bug 75424, I keep *whitespace/declaration* exception for EWebLauncher and MiniBrowser.
> 
> Another (minor) issue here is that this does not go to the .rodata section in the binary; use static const char foo[] instead.

Wow, Thanks.
Fixed like you mentioned.

[Raphael Kubo da Costa (:rakuco)]

As briefly discussed on IRC a few minutes ago, the patch isn't quite ready because of the WK* calls it makes. MiniBrowser should be fairly self-contained in the sense that users should base their code on it and build it outside the tree. That's not possible with the current WK* stuff.

[Ryuan Choi]

Comment on attachment 120903 [details]
Patch

clear flags until efl styled API were prepared.

[Ryuan Choi]

Created attachment 137470 [details]
Patch

[Ryuan Choi]

Created attachment 137497 [details]
Patch

[Dominik Röttsches (drott)]

(In reply to comment #38)
> As briefly discussed on IRC a few minutes ago, the patch isn't quite ready because of the WK* calls it makes. MiniBrowser should be fairly self-contained in the sense that users should base their code on it and build it outside the tree. That's not possible with the current WK* stuff.

Raphael, do you see all your comments addressed in the most recent version of Ryuan's patch? Let's get this in.

[Raphael Kubo da Costa (:rakuco)]

I thought the required WK2 APIs had not been added yet? The patches are passing the EWS bot because we don't enable WebKit2 by default, IIRC (or at least didn't when the last patch was uploaded).

[Gyuyoung Kim]

(In reply to comment #43)
> I thought the required WK2 APIs had not been added yet? The patches are passing the EWS bot because we don't enable WebKit2 by default, IIRC (or at least didn't when the last patch was uploaded).

As far as I know, this patch is not built on EWS yet. Ryuan is submitting patches to support this patch. It looks this patch is able to be reviewed after landing some WK2 patches.

[Ryuan Choi]

Created attachment 145521 [details]
rebased

[Ryuan Choi]

rebased , but Bug 84124 should be landed before this.

[Gyuyoung Kim]

Comment on attachment 145521 [details]
rebased

It looks there are no critical issues as initial implementation. Looks fine.

[Chris Dumez]

Comment on attachment 145521 [details]
rebased

View in context: https://bugs.webkit.org/attachment.cgi?id=145521&action=review

> Tools/MiniBrowser/efl/main.c:116
> +    free(browser);

You never free browser->ee. You probably need a "ecore_evas_free(browser->ee);" before the "free(browser);", right?

[Ryuan Choi]

(In reply to comment #48)
> (From update of attachment 145521 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=145521&action=review
> 
> > Tools/MiniBrowser/efl/main.c:116
> > +    free(browser);
> 
> You never free browser->ee. You probably need a "ecore_evas_free(browser->ee);" before the "free(browser);", right?


ecore_evas_shutdown will release all allocated ecore_evas although we don't call ecore_evas_free explicitly.

If you want to call it explicitly, I will add it.

[Ryuan Choi]

Created attachment 146737 [details]
Patch

[Ryuan Choi]

Created attachment 146986 [details]
rebased to catch r120024

[Gyuyoung Kim]

Comment on attachment 146986 [details]
rebased to catch r120024

View in context: https://bugs.webkit.org/attachment.cgi?id=146986&action=review

> Tools/MiniBrowser/efl/CMakeLists.txt:62
> +SET_TARGET_PROPERTIES(DumpRenderTree PROPERTIES FOLDER "Tools")

By the way, why do you set target folder for DumpRenderTree? Is this correct ?

[Ryuan Choi]

Created attachment 147008 [details]
Fix mistake

[Ryuan Choi]

(In reply to comment #52)
> (From update of attachment 146986 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=146986&action=review
> 
> > Tools/MiniBrowser/efl/CMakeLists.txt:62
> > +SET_TARGET_PROPERTIES(DumpRenderTree PROPERTIES FOLDER "Tools")
> 
> By the way, why do you set target folder for DumpRenderTree? Is this correct ?

Oops, sorry I mistake it while catching up.
Fixed.

[Chang Shu]

Comment on attachment 147008 [details]
Fix mistake

Nice!

[WebKit Review Bot]

Comment on attachment 147008 [details]
Fix mistake

Clearing flags on attachment: 147008

Committed r120130: <http://trac.webkit.org/changeset/120130>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.