Luc R. | 24 Jan 04:41 2015
Picon

Mobile support in WinCairo port

I've been trying to run a few examples of mobile web pages using WinCairo port, and often seeing that content is not re-sized to fit the window like on a smartphone screen. I am setting the window size to be the analog of mobile viewport size (e.g. 320x568) + adding mobile user agent. 
In this setup seems like viewport html tag is not honored and the default behavior is different from a mobile device. Could anyone advice on the architecture of how device-specific rendering is handled in WebKit, should it be implemented on per-port basis, and WinCairo missing this, or am I just missing something obvious?

--
-Luc
<div><div dir="ltr">I've been trying to run a few examples of mobile web pages using WinCairo port, and often seeing that content is not re-sized to fit the window like on a smartphone screen. I am setting the window size to be the analog of mobile viewport size (e.g. 320x568) + adding mobile user agent.&nbsp;<div>In this setup seems like viewport html tag is not honored and the default behavior is different from a mobile device. Could anyone advice on the architecture of how device-specific rendering is handled in WebKit, should it be implemented on per-port basis, and WinCairo missing this, or am I just missing something obvious?<div>
<div><br></div>-- <br><div class="gmail_signature">-Luc</div>
</div>
</div>
</div></div>
Ryan Molden | 20 Jan 19:35 2015
Picon

Errors building unmodified trunk on Windows

Hello, I am trying to get Webkit trunk to build on Windows (using Visual Studio 2013) and running into a few problems.

 

1: I used Git to grab the sources tree, but some parts of the build process, specifically autoversion.pl, seem to assume I used svn, as it does things like svn info (which fails with svn: E155007: '/home/rmolden/WebKit' is not a working copy) to initialize SVN_VERSION and then ends up with

 

Use of uninitialized value $SVN_REVISION in scalar chomp at c:\cygwin\home\rmolden\WebKit\WebKitLibraries\win\tools\scripts\auto-version.pl line 86.

Use of uninitialized value $SVN_REVISION in concatenation (.) or string at c:\cygwin\home\rmolden\WebKit\WebKitLibraries\win\tools\scripts\auto-version.pl line90.

 

    The build then seems to fail because the cmd shell exits with code 1 (which msbuild seems to interpret as failure). Is there any way to get the build working if I used Git to enlist instead of SVN? Or should I just re-enlist using SVN? Or is it failing due to other (unclear) reasons and this is just an innocuous warning?

 

 

2:  I am getting

 

Error      1              error C3646: '__AVAILABILITY_INTERNAL__MAC_10_0_DEP__MAC_10_7' : unknown override specifier (..\platform\sql\SQLiteAuthorizer.cpp)                c:\cygwin\home\rmolden\WebKit\WebKitLibraries\win\include\sqlite3.h           4675       1                WebCore

 

     Hard to tell what its upset about since it appears it has something to do with the macro expansion resulting in an unknown override specifier, but of course it shows the original macro not the expansion in the error message so I am not sure what is ending up tacked on that it is unhappy about L It appears to define these macros in WebKitAvailability, which uses __attribute__, which MSVC certainly doesn’t support.

 

 

3: I am building the WinCairo project, though I can’t really find a good explanation anywhere of what that is. I tried building the normal (debug/release) projects but it gives a ton of errors about missing CoreFoundation stuff, so I presume WinCairo is what I want for Webkit on Windows, but a pointer to something explaining what it is would be appreciated.

 

Thanks in advance for any help you can offer!

 

Ryan

<div>
<div class="WordSection1">
<p class="MsoNormal">Hello, I am trying to get Webkit trunk to build on Windows (using Visual Studio 2013) and running into a few problems.<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">1: I used Git to grab the sources tree, but some parts of the build process, specifically autoversion.pl, seem to assume I used svn, as it does things like svn info (which fails with svn: E155007: '/home/rmolden/WebKit' is not a working
 copy) to initialize SVN_VERSION and then ends up with<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">Use of uninitialized value $SVN_REVISION in scalar chomp at c:\cygwin\home\rmolden\WebKit\WebKitLibraries\win\tools\scripts\auto-version.pl line 86.<p></p></p>
<p class="MsoNormal">Use of uninitialized value $SVN_REVISION in concatenation (.) or string at c:\cygwin\home\rmolden\WebKit\WebKitLibraries\win\tools\scripts\auto-version.pl line90.<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">&nbsp;&nbsp;&nbsp; The build then seems to fail because the cmd shell exits with code 1 (which msbuild seems to interpret as failure). Is there any way to get the build working if I used Git to enlist instead of SVN? Or should I just re-enlist using SVN?
 Or is it failing due to other (unclear) reasons and this is just an innocuous warning?<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">2:&nbsp; I am getting <p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">Error&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 1&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; error C3646: '__AVAILABILITY_INTERNAL__MAC_10_0_DEP__MAC_10_7' : unknown override specifier (..\platform\sql\SQLiteAuthorizer.cpp)&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; c:\cygwin\home\rmolden\WebKit\WebKitLibraries\win\include\sqlite3.h&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
 4675&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 1&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; WebCore<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">&nbsp;&nbsp;&nbsp; &nbsp;Hard to tell what its upset about since it appears it has something to do with the macro expansion resulting in an unknown override specifier, but of course it shows the original macro not the expansion in the error message so I am
 not sure what is ending up tacked on that it is unhappy about <span>
L</span> It appears to define these macros in WebKitAvailability, which uses __attribute__, which MSVC certainly doesn&rsquo;t support.<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">3: I am building the WinCairo project, though I can&rsquo;t really find a good explanation anywhere of what that is. I tried building the normal (debug/release) projects but it gives a ton of errors about missing CoreFoundation stuff, so I presume
 WinCairo is what I want for Webkit on Windows, but a pointer to something explaining what it is would be appreciated.<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">Thanks in advance for any help you can offer!<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">Ryan<p></p></p>
</div>
</div>
Eugene But | 8 Jan 23:33 2015

"build-webkit --simulator" fails

When building Apple's port revision on Mac 178114 using " Tools/Scripts/build-webkit --debug --simulator", I get the following error:

target specifies product type 'com.apple.product-type.tool', but there's no such product type for the 'iphonesimulator' platform

You can find the complete output here: http://pastebin.com/vJfEYVtL

Is is expected that WebKit can not be built for iOS?

Thanks,
Eugene

<div><div dir="ltr">When building Apple's port revision on Mac 178114 using " Tools/Scripts/build-webkit --debug --simulator", I get the following error:<br><br>target specifies product type 'com.apple.product-type.tool', but there's no such product type for the 'iphonesimulator' platform<br><br>You can find the complete output here: <a href="http://pastebin.com/vJfEYVtL">http://pastebin.com/vJfEYVtL</a><div><br></div>
<div>Is is expected that WebKit can not be built for iOS?</div>
<div><br></div>
<div>Thanks,</div>
<div>Eugene</div>
<div><br></div>
</div></div>
Jiho Choi | 29 Dec 16:58 2014
Picon

Overhead of parsing and bytecode generation

Hi all,

I was wondering about the overhead of parsing and bytecode generation. I found the link below saying that it is only ~2% of JS execution.


I got some questions about it.

1. Does it still hold after more than 5 years?
2. I assume that the in-memory code cache was added to reduce the overhead. Does it help across page loads?
3. Is there any plan to make the code cache persistent by writing it to the disk? 

Thanks,
Jiho
<div><div dir="ltr">Hi all,<div><br></div>
<div>I was wondering about the overhead of parsing and bytecode generation. I found the link below saying that it is only ~2% of JS execution.</div>
<div><br></div>
<div>
<a href="http://comments.gmane.org/gmane.os.opendarwin.webkit.devel/7639">http://comments.gmane.org/gmane.os.opendarwin.webkit.devel/7639</a><br>
</div>
<div><br></div>
<div>I got some questions about it.</div>
<div><br></div>
<div>1. Does it still hold after more than 5 years?</div>
<div>2. I assume that the in-memory code cache was added to reduce the overhead. Does it help across page loads?</div>
<div>3. Is there any plan to make the code cache persistent by writing it to the disk?&nbsp;</div>
<div><br></div>
<div>Thanks,</div>
<div>Jiho</div>
</div></div>
Jiho Choi | 10 Dec 20:40 2014
Picon

Is there a documentation for bytecodes?

Hi all,
I was wondering if there is any documentation for JSC's bytecode instructions.
Source/JavaScriptCore/bytecode/BytecodeList.json provides the list with an opcode and the length of each bytecode instruction, and I was wondering if there exists further explanation.

Thanks,
Jiho

<div><div dir="ltr">
<div>Hi all,</div>I was wondering if there is any documentation for JSC's bytecode instructions.<div>Source/JavaScriptCore/bytecode/BytecodeList.json provides the list with an opcode and the length of each bytecode instruction, and I was wondering if there exists further explanation.<br><div><br></div>
<div>Thanks,</div>
<div>Jiho<br><div><br></div>
</div>
</div>
</div></div>
Malek Musleh | 13 Nov 19:20 2014
Picon

Running JetStream Benchmark StandAlone

Hi,

I am not sure if this is the most appropriate mailing list for asking
questions about JetStream.

I want to be able to run the jetstream benchmark in an "offline"
manner, and wanted to confirm if the following methodology is correct
or not.

It looks like there isnt a single tarball or repo for just the
jetstream suite, but I suspect its part of the main WebKit svn repo --
correct?

http://browserbench.org/JetStream/in-depth.html

This link describes the benchmark suite in more detail, and has links
to the specific .js files used.

If I download the corresponding .js files, and then create an
index.html file that "onload" activates the corresponding js script
benchmark. I would then use a browser (e.g. firefox) to point to the
local index.html file, and when it does it should load the js
benchmark.

Is that correct?

For example, I see that the 3d-cube.js file
http://browserbench.org/JetStream/sources/3d-cube.js

contains the same code as the source page of:
http://www.speich.net/projects/programming/3d.htm

and the html page contains a call to Init().

Secondly, I want to know how (if) WebKit invokes the JetStream suite.
Is there a set of regression tests that I can be directed to?

Thanks,

Malek
Dean Jackson | 7 Oct 21:34 2014
Picon

Re: Need help to find Web Messaging and Push-api code


> On 7 Oct 2014, at 5:36 pm, SHIVAKUMAR JAGALUR MATT
<shiva.jm@...> wrote:
> 
> Dear  Jackson,
> 
>         Thanks for the information.  Any reason behind not to add push-api suppourt in webkit ?. It seems W3C and
IETF are now publishing latest push-api spec https://w3c.github.io/push-api/, so can we consider to
add these push-api suppourt in webkit also?

I don't think anyone has spoken out against putting support for push into WebKit.

You (or someone) will need to make a proposal to this list, explaining why you think it's worth having. The
fact that a specification exists isn't enough. We'd also need to have an idea of who will write and maintain
the code, the stability of the technology, what other engines are planning to do, etc.

Dean

> 
> Thanks
> Shiva
> 
> ------- Original Message -------
> Sender : Dean Jackson<dino@...>
> Date : Oct 04, 2014 04:32 (GMT+09:00)
> Title : Re: [webkit-help] Need help to find Web Messaging and Push-api code
> 
> 
> 
> On 30 Sep 2014, at 3:31 pm, SHIVAKUMAR JAGALUR MATT
<shiva.jm@...> wrote:
> 
> 
>            I was trying to find the IDL file and implementation support for Web Messaging
(http://www.w3.org/TR/webmessaging/ ) and Push-api (http://www.w3.org/TR/push-api/ ) in webkit.  I
found the MessagePort.idl, MessagePort.cpp under source/webcore/dom folder are these right files for
Web Messaging?.  But I could not find any idl files for Push-api, is Webkit supports Push-api or no?.
> 
> 
> No, it does not.
> 
> 
> Dean

SHIVAKUMAR JAGALUR MATT | 30 Sep 07:31 2014

Need help to find Web Messaging and Push-api code

Dear   All,

 

             I was trying to find the IDL file and implementation support for Web Messaging (http://www.w3.org/TR/webmessaging/ ) and Push-api (http://www.w3.org/TR/push-api/ ) in webkit.  I found the MessagePort.idl, MessagePort.cpp under source/webcore/dom folder are these right files for Web Messaging?.  But I could not find any idl files for Push-api, is Webkit supports Push-api or no?.

 

Thanks

Shiva

<div><div class="WordSection1">
<p class="MsoNormal">Dear &nbsp;&nbsp;All,<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; I was trying to find the IDL file and implementation support for Web Messaging (<a href="http://www.w3.org/TR/webmessaging/">http://www.w3.org/TR/webmessaging/</a> ) and Push-api (<a href="http://www.w3.org/TR/push-api/">http://www.w3.org/TR/push-api/</a> ) in webkit. &nbsp;I found the MessagePort.idl, MessagePort.cpp under source/webcore/dom folder are these right files for Web Messaging?. &nbsp;But I could not find any idl files for Push-api, is Webkit supports Push-api or no?.<p></p></p>
<p class="MsoNormal"><p>&nbsp;</p></p>
<p class="MsoNormal">Thanks<p></p></p>
<p class="MsoNormal">Shiva<p></p></p>
</div></div>
Alex Shore | 22 Sep 04:29 2014
Picon
Picon

webkit changes after 10.7.5 regarding fixed font default action in filemaker pro versus Safari and default commands to remedy the discrepancy

Hi,

I run a custom database in Filemaker Pro (and Advanced) 11 and 13 that works fine under 10.7.5 and on Windows
7, but once I upgraded to 10.9.4 - the formatting of the webpages filemaker pro (FMP) was displaying was all
distorted.  Apparently this is well known behavior since the webkit changed a default handling of fixed
font family preferences after Lion.

So, in Safari, I can fix the behavior so that displaying the web page in Safari under 10.9.4 works just fine by
running the following default commands:

defaults write com.apple.Safari
com.apple.Safari.ContentPageGroupIdentifier.WebKit2FixedFontFamily Monaco

defaults write com.apple.Safari
com.apple.Safari.ContentPageGroupIdentifier.WebKit2DefaultFixedFontSize 11

I've asked FMP Tech support about it, and that didn't get far; just a "we're not Safari - and where did you get
the idea that the webviewer element should be used to view web pages"

So, if the same FMP database works fine on 10.7.5, but doesn't on 10.9.4 = but viewing the same URL through
Safari on the same machine at the same time after setting the defaults as above renders the page correctly -
I would hope that there is a defaults option that I can set universally for webkit so that when FMP calls it,
it will use the same fixedfontfamily and fixedfondsize as I've told Safari to use.

I tried to explain to the FMP tech support contact that I needed the bundle and Identifier (corresponding to
ContentPageGroupIdentifier) for FMP to write the correct setting - but that didn't get far.

So, short of trying to come up to speed on building a custom version of webkit to run on my machine under 10.9.4
just so I can keep moving forward with FMP and MacOS - is there either a defaults command that I can use to
alter the behavior of FMP?

Or, if there a defaults command(s) that I can use that alters the behavior of webkit for all applications?

Your assistance is appreciated in advance.

My background is way - back (designed EGA and VGA chipsets at IBM back in early 80's) and then designed chips
for HP - but due to math of low noise amplifiers being similar to modern portfolio theory risk calculation,
I do portfolio managment - hence what the database in FMP is used for.  All that is to say I'm real comfortable
with terminal and shells in unix - but not all that familiar with the webdesign tools and terminology. 
Eventually, after burrowing through about 100 levels of code, it still is just poly over silicon - you may
lose me in the terminology of the intermediate levels.

Thank you,
Alex
Koen Bok | 4 Sep 14:30 2014

[Mac Desktop] Show Inspector With Code

Since the patch to allow developer extras landed:

I can set the _setDeveloperExtrasEnabled to get the "Inspect Element" menu on a right mouse click. It then nicely opens a new inspector window in my app.

I would like to show the inspector with a button in my UI, so I was looking how to do that. Can anyone help?

Thanks, Koen
<div><div dir="ltr">Since the patch to allow developer extras landed:<div>
<a href="https://bugs.webkit.org/show_bug.cgi?id=135811">https://bugs.webkit.org/show_bug.cgi?id=135811</a><br>
</div>
<div><br></div>
<div>I can set the&nbsp;<span>_setDeveloperExtrasEnabled to get the "Inspect Element" menu on a right mouse click. It then nicely opens a new inspector window in my app.</span>
</div>
<div><br></div>
<div><span>I would like to show the inspector with a button in my UI, so I was looking how to do that. Can anyone help?</span></div>
<div><span><br></span></div>
<div><span>Thanks, Koen</span></div>
</div></div>
pulkomandy | 2 Sep 15:42 2014

Rendering problems without 3D_RENDERING and TEXTURE_MAPPER

Hi,

I have a rather annoying issue with the Haiku port for some time. In
several places there are pages where parts are not rendered at all. For
example, on youtube.com the search bar at the top of the page does not
render.

I have identified the problem: anything using
	-webkit-transform: translate3d
or any other way to trigger the "3D" mode (even with an identity
transformation) is not rendered.

This would not be a problem if this was used only when 3D is actually
needed, but a lot of websites are using this in an attempt to enable
GPU-accelerated rendering on some browsers.

I tried enabling the 3D_RENDERING build feature but this has no effect
on its own. It seems 3D_RENDERING has an effect only when TEXTURE_MAPPER
is also enabled. We don't want to enable this, for performance reasons
(the texture mapper doesn't work very well with our currently software
rendered drawing code).

I don't think actual 3D rendering is an useful feature for the Haiku
port (at least not until we can use the GPU to do the rendering). So I'm
looking for either:
- A way to make WebKit ignore "translate3d" and the other 3D things in
  CSS and render pages as if there was no transform at all. I thought
  the 3D_RENDERING feature would do that, but apparently it doesn't.
- Or, a way to handle the 3D elements in our platform specific drawing
  code. At the moment it looks like the GraphicsContext mehods are never
  called to draw these 3D elements, but I don't understand why it is so.
  I suspect they are somehow intercepted and sent to a separate "3D"
  drawing code, but I couldn't find that anywhere in the source. So I
  have no idea how they are supposed to be drawn on screen, if they
  don't go through the GraphicsContext.

--

-- 
Adrien.

Gmane