[webkit-dev] browser cant visit the websit

2018-05-21 Thread tugouxp 
hi folks:


   i meet some issue on webkit, when test the midori browser about the vist 
websit function, it doses not work, the network is fine because the curl, wget 
can work properly.  it seems
the issue exist only when visit with the webkit backend. i capture the http 
interaction with "curl" and "midori", it seems different on somewhere.


curl works fine, curl "midori-browser.org"
10.0080:e8:6f:d6:31:d8AtherosC_50:00:01ARP60Who has 192.168.1.253?  Tell 
10.128.128.128
20.78AtherosC_50:00:0180:e8:6f:d6:31:d8ARP42192.168.1.253 is at 
00:03:7f:50:00:01
320.757841192.168.1.253192.168.7.2DNS78Standard query 0x97a7  A 
midori-browser.org
420.767917192.168.7.2192.168.1.253DNS94Standard query response 0x97a7  A 
80.237.133.128
520.769871192.168.1.25380.237.133.128TCP7454498→80 [SYN] Seq=0 Win=29200 Len=0 
MSS=1460 SACK_PERM=1 TSval=379180 TSecr=0 WS=128

6  .


but the midori  is as follows:
10.00192.168.1.253192.168.7.2DNS78Standard query 0x5fdc  A 
midori-browser.org
20.000147192.168.1.253192.168.7.2DNS78Standard query 0x631d   
midori-browser.org
30.013937192.168.7.2192.168.1.253DNS94Standard query response 0x5fdc  A 
80.237.133.128
40.014253192.168.7.2192.168.1.253DNS106Standard query response 0x631d   
2a01:488:42:1000:50ed:8580:2e:30ef
513.89878580:e8:6f:d6:31:d8AtherosC_50:00:01ARP60Who has 192.168.1.253?  Tell 
10.128.128.128
613.898855AtherosC_50:00:0180:e8:6f:d6:31:d8ARP42192.168.1.253 is at 
00:03:7f:50:00:01


it seems there is something different on dns interaction, the midori hold twice 
but the curl only once, the midori seems use the ipv6 address, is this the 
rootcause?


 thanks for your kindly support!___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] Request for upgrading xcode toolchains in Apple bots for C++17

2018-05-21 Thread Yusuke SUZUKI 
Thanks,

On Tue, May 22, 2018 at 1:30 AM Alexey Proskuryakov  wrote:

> Hi,
>
> To clarify, are you asking specifically for mac-32bit EWS to be upgraded?
>

No, I'm asking for upgrading mac/ios bots except for mac-32bit.


> It has Xcode 9.2, and can be upgraded to 9.3.1 (that will be a multi-step
> process, as we need to keep the version in sync with several other setups).
> But macOS Sierra bots have Xcode 8.3.3, and that is the latest release for
> Sierra.
>

Oh! Hmm, OK. So, to avoid build failures, we need to have the way to
distinguish these 8.3.3's clang from 9.2 - 9.3.1 clangs.
In terms of the implementation status of C++17, it is not a problem. Even
clang 4 shows good status compared to GCC 6.


>
> - Alexey
>
>
> 21 мая 2018 г., в 6:44, Yusuke SUZUKI  написал(а):
>
> Hi WebKittens, in particular Apple bot maintainers!
>
> We are about to enable C++17 in all the WebKit ports, and the last step is
> enabling C++17 for Xcode[1].
> While the latest shipped Xcode (w/ clang) supports C++17 option
> (`-std=gnu++17`), EWS build bots do not support this. Some build bots
> accept `-std=c++1z`, this option causes trouble in the latest Xcode (e.g.
> mac-32bit uses the newer Xcode).
>
> Can we upgrade Xcodes in EWS and buildbots to the latest ones to accept
> C++17 option? Once it is done, WebKit starts using fancy C++17 features
> listed in GCC 6.0.
>
> Best regards,
> Yusuke Suzuki
>
> [1]: https://bugs.webkit.org/show_bug.cgi?id=185176
>
> ___
> webkit-dev mailing list
> webkit-dev@lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev
>
>
> - Alexey
>
>
>
>
___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] Request for upgrading xcode toolchains in Apple bots for C++17

2018-05-21 Thread Alexey Proskuryakov 
Hi,

To clarify, are you asking specifically for mac-32bit EWS to be upgraded? It 
has Xcode 9.2, and can be upgraded to 9.3.1 (that will be a multi-step process, 
as we need to keep the version in sync with several other setups). But macOS 
Sierra bots have Xcode 8.3.3, and that is the latest release for Sierra.

- Alexey


> 21 мая 2018 г., в 6:44, Yusuke SUZUKI  написал(а):
> 
> Hi WebKittens, in particular Apple bot maintainers!
> 
> We are about to enable C++17 in all the WebKit ports, and the last step is 
> enabling C++17 for Xcode[1].
> While the latest shipped Xcode (w/ clang) supports C++17 option 
> (`-std=gnu++17`), EWS build bots do not support this. Some build bots accept 
> `-std=c++1z`, this option causes trouble in the latest Xcode (e.g. mac-32bit 
> uses the newer Xcode).
> 
> Can we upgrade Xcodes in EWS and buildbots to the latest ones to accept C++17 
> option? Once it is done, WebKit starts using fancy C++17 features listed in 
> GCC 6.0.
> 
> Best regards,
> Yusuke Suzuki
> 
> [1]: https://bugs.webkit.org/show_bug.cgi?id=185176 
> 
> 
> ___
> webkit-dev mailing list
> webkit-dev@lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev

- Alexey



___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] [jsc-dev] Proposal: Removing ENABLE(INTL)

2018-05-21 Thread Michael Catanzaro 
On Mon, May 21, 2018 at 9:45 AM, Michael Catanzaro 
 wrote:
But I'm curious if anyone knows what the ICU requirement is for 
ENABLE(INTL)?


Also, Ms2ger reported https://bugs.webkit.org/show_bug.cgi?id=185714 
recently. It seems there are new options 
ENABLE_INTL_NUMBER_FORMAT_TO_PARTS and ENABLE_INTL_PLURAL_RULES, both 
disabled in WebKitFeatures.cmake. Ideally those would be enabled if ICU 
is new enough. Does anybody know what ICU versions are required for 
these? (If so, best comment in the bug.)


___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] [jsc-dev] Proposal: Removing ENABLE(INTL)

2018-05-21 Thread Konstantin Tokarev 


21.05.2018, 17:45, "Michael Catanzaro" :
> On Mon, May 21, 2018 at 8:55 AM, Konstantin Tokarev 
> wrote:
>>  For the record, having !ENABLE(INTL) option allowed to build WebKit
>>  against old ICU versions, e.g. version that is shipped with RHEL 7.
>
> For WebKitGTK+, we do not support building without ENABLE(INTL), and we
> don't need to worry about RHEL 7 since RHEL is shipping an unsupported
> version of WebKit with bundled ICU.
>
> But we do need to support Ubuntu 16.04 (with ICU 55). If ENABLE(INTL)
> is going to require newer ICU than that, then we would need to expose
> ENABLE(INTL) and ensure that disabling it works. We do have a buildbot
> to check this, and it's green right now, so trunk must be fine. But I'm
> curious if anyone knows what the ICU requirement is for ENABLE(INTL)?

AFAIK ICU 54 is enough

>
> Michael

-- 
Regards,
Konstantin

___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] [jsc-dev] Proposal: Removing ENABLE(INTL)

2018-05-21 Thread Michael Catanzaro 
On Mon, May 21, 2018 at 8:55 AM, Konstantin Tokarev  
wrote:
For the record, having !ENABLE(INTL) option allowed to build WebKit 
against old ICU versions, e.g. version that is shipped with RHEL 7.


For WebKitGTK+, we do not support building without ENABLE(INTL), and we 
don't need to worry about RHEL 7 since RHEL is shipping an unsupported 
version of WebKit with bundled ICU.


But we do need to support Ubuntu 16.04 (with ICU 55). If ENABLE(INTL) 
is going to require newer ICU than that, then we would need to expose 
ENABLE(INTL) and ensure that disabling it works. We do have a buildbot 
to check this, and it's green right now, so trunk must be fine. But I'm 
curious if anyone knows what the ICU requirement is for ENABLE(INTL)?


Michael

___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

Re: [webkit-dev] [jsc-dev] Proposal: Removing ENABLE(INTL)

2018-05-21 Thread Konstantin Tokarev 


19.05.2018, 14:12, "Yusuke SUZUKI" :
> Hi WebKittens,
>
> I would like to remove ENABLE(INTL) compile time flag and always enable INTL.
> This is ECMAScript Intl (i18n) feature. This feature depends on ICU.
> But right now, even JSC (and WTF) always require ICU. Even if we disable 
> ENABLE(INTL), we still require ICU.

For the record, having !ENABLE(INTL) option allowed to build WebKit against old 
ICU versions, e.g. version that is shipped with RHEL 7.

>
> New INTL features are gurded separately with the compile flags (since some 
> new features require new ICU). So, removing ENABLE(INTL) just enables Intl 
> basic features, which are shipped more than 1 year ago and it is quite mature.
>
> Best regards,
> Yusuke Suzuki
> --
> Regards,
> Yusuke Suzuki
> ,
>
> ___
> jsc-dev mailing list
> jsc-...@lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/jsc-dev


-- 
Regards,
Konstantin
___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev

[webkit-dev] Request for upgrading xcode toolchains in Apple bots for C++17

2018-05-21 Thread Yusuke SUZUKI 
Hi WebKittens, in particular Apple bot maintainers!

We are about to enable C++17 in all the WebKit ports, and the last step is
enabling C++17 for Xcode[1].
While the latest shipped Xcode (w/ clang) supports C++17 option
(`-std=gnu++17`), EWS build bots do not support this. Some build bots
accept `-std=c++1z`, this option causes trouble in the latest Xcode (e.g.
mac-32bit uses the newer Xcode).

Can we upgrade Xcodes in EWS and buildbots to the latest ones to accept
C++17 option? Once it is done, WebKit starts using fancy C++17 features
listed in GCC 6.0.

Best regards,
Yusuke Suzuki

[1]: https://bugs.webkit.org/show_bug.cgi?id=185176
___
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev