[webkit-gtk] Is there any OSX pre-build library?
Hi everyone, I'm using `sudo port install webkit-gtk` as mentioned in the official page and it takes just about forever compared with any other Linux distro, I wonder if there's any pre-build shared library that could work? libwebkit-gtk in Ubuntu works quite well and so does webkitgtk in ArchLinux. Last, but not least, Homebrew formulas found here and there seem to be broken. Thanks for any help/hint/info about this. Best Regards ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK+ 2.12.0 released!
It looks like it's impossible to build it on OSX since, at least, version 2.11.92: https://github.com/Homebrew/homebrew/pull/50565#issuecomment-204692495 Used formula and it args are here: https://github.com/WebReflection/homebrew-gnome/blob/master/webkitgtk.rb#L22-L37 Does anybody know if there's a problem with the gir scanner or if it might be the used formula? Thank You On Tue, Mar 22, 2016 at 9:53 AM, Carlos Garcia Camposwrote: > WebKitGTK+ 2.12.0 is available for download at: > > http://webkitgtk.org/releases/webkitgtk-2.12.0.tar.xz (11.1MB) >md5sum: d233b5594c37a961308c230207d8b64f >sha256sum: > a68604f8fd1bcb247d647709290226289b90c16acb1f3730538b688db8df5ea6 > > This is the first stable release in the 2.12 series. > > Highlights of the WebKitGTK+ 2.12.0 release > === > > - Enable FTL by default in JavaScriptCore for x86_64. > - Network process is now used unconditionally. The shared secondary > process model is now > the same as using the multiple process model and setting a process > limit of 1. > - Switch to use overlay scrollbars like all other GTK+ widgets and > ensure the behavior > is consistent with GTK+ too. > - Support for windowless NPAPI plugins with no UI in non X11 platforms. > - Enable GSS-Negotiate support when available in libsoup. > - Improved general performance by better handling glib main loop sources. > - New API to save and restore a WebView session. > > > What is WebKitGTK+? > === > > WebKitGTK+ is the GNOME platform port of the WebKit rendering engine. > Offering WebKit's full functionality through a set of GObject-based > APIs, it is suitable for projects requiring any kind of web > integration, from hybrid HTML/CSS applications to full-fledged web > browsers. > > Requirements > > > gtk+ >= 3.6.0 > glib >= 2.36 > libsoup >= 2.42.0 > cairo >= 1.10.2 > libxml >= 2.8.0 > libxslt >= 1.1.7 > HarfBuzz >= 0.9.2 > icu > fontconfig >= 2.8.0 > FreeType2 >= 2.4.2 > libsecret > SQLite > enchant > > Depending on your configuration options WebKitGTK+ may also depend on: > > gtk+ >= 2.24.10 > GObject introspection > gstreamer >= 1.0.3 > GeoClue >= 2.1.5 > gudev > cairo-gl >= 1.10.2 > > More information > > > If you want to know more about the project or get in touch with us > you may: > > - Visit our website at http://www.webkitgtk.org, or the upstream > site at http://www.webkit.org. People interested in contributing > should read: http://www.webkit.org/coding/contributing.html. > > - Browse the bug list at http://bugs.webkit.org, WebKitGTK+ bugs are > typically prefixed by "[GTK]." A bug report with a minimal, > reproducible test case is often just as valuable as a patch. > > - Join the #webkit and #webkitgtk+ IRC channels at irc.freenode.net. > > - Subscribe to the WebKitGTK+ mailing list, > http://lists.webkit.org/mailman/listinfo.cgi/webkit-gtk, or the > WebKit development mailing list, > http://lists.webkit.org/mailman/listinfo.cgi/webkit-dev > > Thanks > == > > Thanks to all the contributors who made possible this release, they > are far too many to list! > > The WebKitGTK+ team, > March 22, 2016 > -- > Carlos Garcia Campos > http://pgp.rediris.es:11371/pks/lookup?op=get=0xF3D322D0EC4582C3 > > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK+ 2.12.0 released!
Thank You! On Sat, Apr 2, 2016 at 7:42 PM, Jeremy Huddleston Sequoia < jerem...@apple.com> wrote: > It's been building in MacPorts for the past few months. We fixed a ton of > OS X issues in webkit-gtk during 2.11.x. As discussed in multiple > locations already, this is a HomeBrew + SIP issue. You can't pass > DYLD_LIBRARY_PATH to system executables; it will be ignored. /usr/bin/env > and /usr/bin/python are system executables. HomeBrew can address this by > using their own python as gir's interpreter. > > --Jeremy > > > On Apr 2, 2016, at 09:00, Andrea Giammarchi <andrea.giammar...@gmail.com> > wrote: > > > > It looks like it's impossible to build it on OSX since, at least, > version 2.11.92: > > https://github.com/Homebrew/homebrew/pull/50565#issuecomment-204692495 > > > > Used formula and it args are here: > > > https://github.com/WebReflection/homebrew-gnome/blob/master/webkitgtk.rb#L22-L37 > > > > Does anybody know if there's a problem with the gir scanner or if it > might be the used formula? > > > > Thank You > > > > > > On Tue, Mar 22, 2016 at 9:53 AM, Carlos Garcia Campos < > cgar...@igalia.com> wrote: > > WebKitGTK+ 2.12.0 is available for download at: > > > > http://webkitgtk.org/releases/webkitgtk-2.12.0.tar.xz (11.1MB) > >md5sum: d233b5594c37a961308c230207d8b64f > >sha256sum: > a68604f8fd1bcb247d647709290226289b90c16acb1f3730538b688db8df5ea6 > > > > This is the first stable release in the 2.12 series. > > > > Highlights of the WebKitGTK+ 2.12.0 release > > === > > > > - Enable FTL by default in JavaScriptCore for x86_64. > > - Network process is now used unconditionally. The shared secondary > process model is now > > the same as using the multiple process model and setting a process > limit of 1. > > - Switch to use overlay scrollbars like all other GTK+ widgets and > ensure the behavior > > is consistent with GTK+ too. > > - Support for windowless NPAPI plugins with no UI in non X11 platforms. > > - Enable GSS-Negotiate support when available in libsoup. > > - Improved general performance by better handling glib main loop > sources. > > - New API to save and restore a WebView session. > > > > > > What is WebKitGTK+? > > === > > > > WebKitGTK+ is the GNOME platform port of the WebKit rendering engine. > > Offering WebKit's full functionality through a set of GObject-based > > APIs, it is suitable for projects requiring any kind of web > > integration, from hybrid HTML/CSS applications to full-fledged web > > browsers. > > > > Requirements > > > > > > gtk+ >= 3.6.0 > > glib >= 2.36 > > libsoup >= 2.42.0 > > cairo >= 1.10.2 > > libxml >= 2.8.0 > > libxslt >= 1.1.7 > > HarfBuzz >= 0.9.2 > > icu > > fontconfig >= 2.8.0 > > FreeType2 >= 2.4.2 > > libsecret > > SQLite > > enchant > > > > Depending on your configuration options WebKitGTK+ may also depend on: > > > > gtk+ >= 2.24.10 > > GObject introspection > > gstreamer >= 1.0.3 > > GeoClue >= 2.1.5 > > gudev > > cairo-gl >= 1.10.2 > > > > More information > > > > > > If you want to know more about the project or get in touch with us > > you may: > > > > - Visit our website at http://www.webkitgtk.org, or the upstream > > site at http://www.webkit.org. People interested in contributing > > should read: http://www.webkit.org/coding/contributing.html. > > > > - Browse the bug list at http://bugs.webkit.org, WebKitGTK+ bugs are > > typically prefixed by "[GTK]." A bug report with a minimal, > > reproducible test case is often just as valuable as a patch. > > > > - Join the #webkit and #webkitgtk+ IRC channels at irc.freenode.net. > > > > - Subscribe to the WebKitGTK+ mailing list, > > http://lists.webkit.org/mailman/listinfo.cgi/webkit-gtk, or the > > WebKit development mailing list, > > http://lists.webkit.org/mailman/listinfo.cgi/webkit-dev > > > > Thanks > > == > > > > Thanks to all the contributors who made possible this release, they > > are far too many to list! > > > > The WebKitGTK+ team, > > March 22, 2016 > > -- > > Carlos Garcia Campos > > http://pgp.rediris.es:11371/pks/lookup?op=get=0xF3D322D0EC4582C3 > > > > > > ___ > > webkit-gtk mailing list > > webkit-gtk@lists.webkit.org > > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > > > > > ___ > > webkit-gtk mailing list > > webkit-gtk@lists.webkit.org > > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] Send HTTP header line for every request
FWIW, we've been exploring and talking about having V8 driven Gtk, instead of Moz one. My humble, and working, attempt to at least have ES2015 syntax and features, and a more NodeJS friendly environment for GJS is here: https://github.com/WebReflection/jsgtk#jsgtk All GJS examples run either out of the box or after making them more JavaScriptish: https://github.com/WebReflection/jsgtk-examples#jsgtk-examples And there is also a Twitter client based on WebkitGTK+ https://github.com/WebReflection/jsgtk-twitter#a-desktop-client-for-twitter-mobile Best Regards On Tue, Jun 28, 2016 at 2:26 PM, Adrián Pérez de Castrowrote: > Quoting Jay Strict (2016-06-28 11:57:31) > > On 28.06.2016 08:05, philip.chime...@gmail.com wrote: > > > I am fairly certain GJS is the more frequently used of the two, in > GNOME > > > applications at least. In any case, GJS is not just used for GNOME > > > Shell, but also several applications like GNOME Documents (which itself > > > uses WebKitGTK), GNOME Weather, etc. > > > > > > And after all, I don't suppose it's a burden to include a dependency on > > > Python for the Python shim. You might be able to write WebExtension > > > shims for any number of interpreted languages and make them an optional > > > part of the build. In package-based distributions, for example, they > > > could be shipped in separate packages. > > > > That would not help in the Gnome shell extension use case then, if you > > have to install a seperate package. Gnome shell extensions should work > > for different distributions no matter which extra packages are installed > > or not. > > I must admit that this is a very good reason to go with GJS instead of > Seed. > Thinking about usage in GNOME, probably it makes more sense to start with > GJS > first, and forget about supporting Seed for now. Maybe I'll try to make a > standalone proof-of-concept shim loader like the Python one using GJS when > I have some time — though there's still a couple of other tasks waiting in > my list of WIP tasks for WebKitGTK+, so I cannot commit to a date, sorry. > > > In an ideal world, GNOME shouldn't need to depend on two separate > JavaScript > engines, and the Shell would use JavaScriptCore somehow (either making GJS > depend on JSC instead of SpiderMonkey, or via Seed). Of course at the time > when GJS appeared SpiderMonkey was chosen —probably I would take that > decision > myself back then, too—, and code in the Shell depends (still?) on some > features which are available in SpiderMonkey only, so now we have to live > with > that. > > I would love to see GJS ported over to JavaScriptCore, to avoid having both > JavaScriptCore and SpiderMonkey working at the same in the same process > for an > application, but I am aware that it is not a motivating project because GJS > “just works right now” and if GJS switched to JSC probably the Shell would > need modifications as well. Not to mention making sure that there are no > regressions, and that then the Shell would depend on WebKitGTK+, whic > itself > could open another can of worms... > > So no, I do not see GJS switching to JSC anytime soon. Unfortunately. > > > Cheers, > > -- > ☛ Adrián > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
[webkit-gtk] WebKitGTK+ VS Apple WebKit
Hello there, I've recently read this announcement: https://twitter.com/webkit/status/823967381026263040 which as summary says: "New @WebKit features in Safari 10.1 with the macOS 10.12.4 & iOS 10.3 betas: CSS Grid, Fetch, and so much more. " I am a GNOME on ArchLinux happy user but Web, which AFAIK is based on WebKit2GTK+, does not have even half the features Safari or WebKit Nightly has on Apple OS. I wonder if there's any browser, even experimental or from AUR, that uses latest features that land on Safari or if such browser has a special fork and development is different from what lands on Web or WebKitGTK based browsers. Thanks for any sort of clarification. Best Regards ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK+ VS Apple WebKit
Couldn't expect a better answer, thanks a lot. If I might ask my last question: who makes the decision of "what ships where"? As example, how comes Safari is shipping customElements but latest WebKitGTK+ doesn't even have them behind a flag? Any specific procedure to speedup shipping features like that? Thanks again and Best Regards. On Wed, Jan 25, 2017 at 2:59 PM, Adrian Perez de Castro <ape...@igalia.com> wrote: > Hello Andrea, > > On Wed, 25 Jan 2017 10:04:43 +, Andrea Giammarchi < > andrea.giammar...@gmail.com> wrote: > > Hello there, > > I've recently read this announcement: > > > > https://twitter.com/webkit/status/823967381026263040 > > > > which as summary says: > > "New @WebKit features in Safari 10.1 with the macOS 10.12.4 & iOS 10.3 > > betas: CSS Grid, Fetch, and so much more. " > > > > I am a GNOME on ArchLinux happy user but Web, which AFAIK is based on > > WebKit2GTK+, does not have even half the features Safari or WebKit > Nightly > > has on Apple OS. > > > > I wonder if there's any browser, even experimental or from AUR, that uses > > latest features that land on Safari or if such browser has a special fork > > and development is different from what lands on Web or WebKitGTK based > > browsers. > > The following is a quick breakdown, without including Safari-specific > features: > > - Working as of WebKitGTK+ 2.14.3 (some of them have been available even in > earlier versions): > > - Fetch API, demo: https://addyosmani.com/demos/fetch-api/ > - IndexedDB, demo: https://robnyman.github.io/ > html5demos/indexeddb/ > - EmcmaScript 2016 & 2017: Whatever JavaScriptCore supports, is > included > in WebKitGTK+, exactly the same as in Apple's ports. > - CSS Grid Layout, demo: http://gridbyexample.com/examples/ > (Actually, the team implementing it works at Igalia with us, and > we have > had it enabled a few versions back already!) > - Improved Web Inspector Debugging. Again, we use the same code as > Apple > for the inspector. > > - Working at least partially, but I have doubts about how the GTK+ handles > specifics: > > - CSS Deep Colors: Again, we use the same CSS support code as > Apple, so > confident the new syntax is parsed even without trying :-D. > Cannot > guarantee that the internal color representation used by our > port can > represent with enough precision. AFAIK Cairo uses floating point > components for representing colors, which is more than enough, > but that > alone may not be enough to output the colors defined in CSS to a > screen > with a wide color gamut. > > - Not working yet, but IMHO likely to be easily doable (e.g. activatable at > build time, without needing additional support in the port): > > - HTML Custom Elements, demo: https://webkit.org/wp-content/ > uploads/custom-element-demo.html > - Input Events, demo: http://jsfiddle.net/girlie_mac/sfxYG/ > - Interactive form validation, demo: http://codepen.io/cdumez/full/ > zoOZmZ/ > (Funnily enough, I see no errors in the inspector console, dunno > why > it wouldn't work.) > > - Not working yet, may need port-specific support code: > > - Pointer Lock, demo: https://mdn.github.io/pointer-lock-demo/ > - Gamepad API, demo: http://html5gamepad.com/ > (IIRC, there was some WIP patches for this being done at some > point.) > - HTML5 Download Attribute, demo: http://cdn.sixrevisions.com/ > 0435-01_html5_download_attribute_demo/html5download-demo.html > - Reduced Motion Media Query: This shouldn't be too difficult, the CSS > parsing is handled by common code, but we lack in the GTK+ port > an API > to tell WebKit whether to use items with the > “prefers-reduced-motion” > CSS media query. > > > Thanks for any sort of clarification. > > Hopefully the list above helps. As you can see, WebKitGTK+ does not fare > badly, and some of the new features may be activatable at build time > without > needing to add new port-specific code. > > Cheers, > > — > Adrián > > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK+ VS Apple WebKit
Thanks, and to confirm I have 2.14 indeed, 2.15 is in AUR as unstable and not brebuilt. I guess at this point I'll wait those two months :-) Best Regards On Wed, Jan 25, 2017 at 3:54 PM, Michael Catanzaro <mcatanz...@igalia.com> wrote: > On Wed, 2017-01-25 at 15:29 +0000, Andrea Giammarchi wrote: > > If I might ask my last question: who makes the decision of "what > > ships where"? > > Whoever is working on the feature. If the people working on the feature > decide to enable it in WebKitGTK+, then we'll get it. If not, we'll > never get it until someone working on WebKitGTK+ decides they want it. > Nobody keeps track of the delta in feature differences between the two > ports. It's certainly something that would be desirable. > > > As example, how comes Safari is shipping customElements but latest > > WebKitGTK+ doesn't even have them behind a flag? > > Ah, I figured this out: the problem is you are testing the stable > branch. In 2.14 there is a hidden and unsupported > -DENABLE_CUSTOM_ELEMENTS build flag you can try using, but it might not > be enough because it'll probably be disabled at runtime as well, and we > have no way to enumerate or enable runtime experimental features except > for CSS grid. We have a WebKitGTK+ experimental features environment > variable that only covers CSS grid and does not map to the actual > experimental features we have in WebKit. We should probably remove this > or fix it somehow because it's surely not working properly right now: > it's not OK for it to only work for only one particular experimental > feature. > > But if you're testing 2.15.3, custom elements should be enabled and > working by default. At least I see a progress bar on the custom element > demo page, exactly the same as in Chromium, whereas in 2.14 I don't see > anything. So the problem is just different release cycles. We really do > need the full six months to make sure we're releasing a stable product. > 2.16 will be released in two months. > > Michael > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] [jsc-dev] Enable WebAssembly on Linux based ports (GTK and JSCOnly)
Chiming in, it'd be awesome to be able to test WASM on 2.16 behind a flag, I'm targeting mostly SBC including Raspberry Pi 0 where WebKitGTK+ already works like a charm (enabling WebGL and with HW accelerated layers) but I can't compare performance against WASM and building from scratch targeting Pi Zero ain't the easiest thing to do. Thanks for eventual consideration. On Wed, Mar 1, 2017 at 4:57 PM, Yusuke SUZUKIwrote: > On Thu, Mar 2, 2017 at 1:36 AM, Saam Barati wrote: > >> >> On Mar 1, 2017, at 7:46 AM, Michael Catanzaro >> wrote: >> >> On Wed, 2017-03-01 at 15:44 +0100, Carlos Garcia Campos wrote: >> >> We have recently branched for 2.16 so it's a perfect moment to enable >> >> it in trunk with very little risk. >> >> >> Is Apple planning to enable it within the next six months? If so, now >> is indeed the ideal time to enable it for WebKitGTK+. >> >> >> What do you mean by enable it? It's been on by default on the Mac port >> for a while now and has been in the various STP releases. >> >> I think it's a good idea to enable it on other ports too. Note that Keith >> is currently working on a patch to make loads faster in Wasm. It may >> require some OS specific handling to run well on Linux: >> https://bugs.webkit.org/show_bug.cgi?id=162693 >> >> > Awesome! I think enabling WASM on Linux after landing Keith's faster > memory access patch (by using SEGV signal handling instead of bound > checking) is good. > Once that patch is implemented and landed, we can start implementing > platform-specific part of that (extracting instruction pointer from > mcontext_t and signal handling part should be implemented for Linux). > I think that should be easy (since retrieving instruction pointer from > mcontext_t is already once implemented for sampling profiler. Thus, we can > share / re-implement the same thing for that). > > >> >> If not, then I'd prefer to wait until it's going to be enabled in the >> macOS port. >> >> Michael >> ___ >> jsc-dev mailing list >> jsc-...@lists.webkit.org >> https://lists.webkit.org/mailman/listinfo/jsc-dev >> >> > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] Modifying HTML before loading it in a WebView
Out of curiosity: wouldn't an injected script in the page enabled via sha-nounce as the only one executable be enough, if CSP support is available? On Fri, Oct 7, 2016 at 10:35 AM, Michael Catanzarowrote: > On Fri, 2016-10-07 at 09:25 +0200, Carlos Garcia Campos wrote: > > That's a very good point, we'll check if that's possible, or wer'll > > try > > to make it possible otherwise. > > Wow, I never considered this. > > Took a quick look at Epiphany. Guess how many Epiphany features > implemented in JS break when you use the hidden disable JS setting. > (Four.) > > Michael > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK+ VS Apple WebKit
FYI the test apparently looks for V0 API which is based on document.registerElement, hence the test fails. I've filed a bug already, I'm sure they'll change that. https://github.com/NielsLeenheer/html5test/issues/501 However, it's surprising the only thing missing thing in WebKitGTK to be a fully ES2015 Standard compliant browser is the module export/import through
Re: [webkit-gtk] WebKitGTK+ VS Apple WebKit
I was hoping from something I could call via PyGTK or GJS to be honest, like neabling WebGL and stuff like that. The part I love the most about WebKitGTK is that it runs even on a Raspberry Pi 0 but I don't want to build for every platform that has been been built by ArchLinux already, I'm not as good as them at configuring build targets. Thanks for any hint, still I'd like to know if ES2015 modules are at least in the pipeline, if possible. Thanks On Tue, Apr 4, 2017 at 9:29 PM, Adrian Perez de Castrowrote: > On Mon, 03 Apr 2017 18:43:10 -0500, Michael Catanzaro < > mcatanz...@igalia.com> wrote: > > > I don't think Arch is missing build flags (in fact I strongly encourage > > using the defaults), [...] > > Arch Linux uses the defaults, as you can see in the CMake invocation used > for > building at: > > https://git.archlinux.org/svntogit/packages.git/tree/ > trunk/PKGBUILD?h=packages/webkit2gtk#n33 > > If you need (or want) to try enabling additional features, you can grab the > “PKGBUILD” from there and add the “-DENABLE_…” flags for the features you > might want to try. Of course, YMMV and all that :-) > > Br. > > -- > Adrián > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK+ VS Apple WebKit
Thanks for the answer, it looks like the version built for ArchLinux is missing some flag then. I'll try to figure out what's going on there. However ... > Our team does not usually prioritize web-facing features I wonder then, how's Gnome Web (Epiphany) trying to be the default Desktop/platform browser ? Beside that: > this webpage is what tech reviewers use to judge the quality of web browsers That page simply uses features detection, what every Web developer in this world would use instead of excluding through UA sniffing anythign that's not Chrome. If CustomElementsRegistry is in there and customElements.define works I don't see why WebKitGTK would fail. I know the author of that page though, and I'm both a Web/NodeJS developer and Web/NodeJS Polyfills author myself so I can counter verify what's going on. Last question, if I might: > Our team does not usually prioritize web-facing features a Web engine that doesn't prioritize web-facing features sounds quite an "unsound" statement. What's the official WebKitGTK focus/priority then? Thank You! On Mon, Apr 3, 2017 at 9:45 PM, Michael Catanzaro <mcatanz...@igalia.com> wrote: > On Mon, 2017-04-03 at 19:18 +0100, Andrea Giammarchi wrote: > > Dear All, > > I've just played a bit around with WebKit2GTK 2.16 for ArchLinux, > > out of Gnome-Unstable repository. > > > > The http://html5test.com/ score is like 8 points more, still under > > 400 out of 555. > > > > Moreover, Custom Elements, as well as ES2015 script type=module are > > not supported. > > > > Am I missing flags during my initialization or nothing recent > > actually went into this version? > > > > Fetch is not there neither. > > > > Mind explaining what's going on? > > > > Thank You! > > Custom elements works for me: > > https://webkit.org/wp-content/uploads/custom-element-demo.html > > The progress bar does nothing in 2.14 but works fine in 2.16. I see > html5test.com scores it as unsupported. I'm curious why. For better or > for worse, this webpage is what tech reviewers use to judge the quality > of web browsers, and we're missing four points for custom elements. > > I dunno about ES6 modules, but I thought that should be working too. > > html5test.com says Fetch is working. > > Our team does not usually prioritize web-facing features except when > somebody is paying us to do so (CSS grid layout), so I'm not at all > familiar with any of the above. Maybe somebody else knows. For an > overview of what we've been working on, see: > > http://blogs.igalia.com/carlosgc/2017/03/20/webkitgtk-2-16/ > > Michael > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] [jsc-dev] Enable WebAssembly on Linux based ports (GTK and JSCOnly)
> AFAIU Zero is based on 32-bit ARM, and WASM requires B3 that is 64-bit only. performance for the already performant, seems legit :D I didn't know this precious detail, thanks a lot for the info! Never mind my previous email. Best Regards On Wed, Mar 1, 2017 at 5:16 PM, Konstantin Tokarev <annu...@yandex.ru> wrote: > > > 01.03.2017, 20:10, "Andrea Giammarchi" <andrea.giammar...@gmail.com>: > > Chiming in, it'd be awesome to be able to test WASM on 2.16 behind a > flag, I'm targeting mostly SBC including Raspberry Pi 0 where WebKitGTK+ > already works like a charm (enabling WebGL and with HW accelerated layers) > but I can't compare performance against WASM and building from scratch > targeting Pi Zero ain't the easiest thing to do. > > AFAIU Zero is based on 32-bit ARM, and WASM requires B3 that is 64-bit > only. > > > > > Thanks for eventual consideration. > > > > On Wed, Mar 1, 2017 at 4:57 PM, Yusuke SUZUKI <utatane@gmail.com> > wrote: > >> On Thu, Mar 2, 2017 at 1:36 AM, Saam Barati <sbar...@apple.com> wrote: > >>> On Mar 1, 2017, at 7:46 AM, Michael Catanzaro <mcatanz...@igalia.com> > wrote: > >>> > >>>> On Wed, 2017-03-01 at 15:44 +0100, Carlos Garcia Campos wrote: > >>>> We have recently branched for 2.16 so it's a perfect moment to enable > >>>> it in trunk with very little risk. > >>>> > >>>> Is Apple planning to enable it within the next six months? If so, now > >>>> is indeed the ideal time to enable it for WebKitGTK+. > >>> > >>> What do you mean by enable it? It's been on by default on the Mac port > for a while now and has been in the various STP releases. > >>> > >>> I think it's a good idea to enable it on other ports too. Note that > Keith is currently working on a patch to make loads faster in Wasm. It may > require some OS specific handling to run well on Linux: > https://bugs.webkit.org/show_bug.cgi?id=162693 > >> > >> Awesome! I think enabling WASM on Linux after landing Keith's faster > memory access patch (by using SEGV signal handling instead of bound > checking) is good. > >> Once that patch is implemented and landed, we can start implementing > platform-specific part of that (extracting instruction pointer from > mcontext_t and signal handling part should be implemented for Linux). > >> I think that should be easy (since retrieving instruction pointer from > mcontext_t is already once implemented for sampling profiler. Thus, we can > share / re-implement the same thing for that). > >> > >>>> If not, then I'd prefer to wait until it's going to be enabled in the > >>>> macOS port. > >>>> > >>>> Michael > >>>> ___ > >>>> jsc-dev mailing list > >>>> jsc-...@lists.webkit.org > >>>> https://lists.webkit.org/mailman/listinfo/jsc-dev > >> > >> ___ > >> webkit-gtk mailing list > >> webkit-gtk@lists.webkit.org > >> https://lists.webkit.org/mailman/listinfo/webkit-gtk > > , > > > > ___ > > webkit-gtk mailing list > > webkit-gtk@lists.webkit.org > > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > > -- > Regards, > Konstantin > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK+ VS Apple WebKit
That'd be awesome news. I'll double check later on and eventually confirm it (if needed). Best Regards On Wed, Apr 5, 2017 at 11:02 AM, Yusuke SUZUKI <utatane@gmail.com> wrote: > I believe ES6 modules are enabled in all the platforms. > > On Wed, Apr 5, 2017 at 6:08 PM, Adrian Perez de Castro <ape...@igalia.com> > wrote: > >> On Tue, 4 Apr 2017 23:47:29 +0100, Andrea Giammarchi < >> andrea.giammar...@gmail.com> wrote: >> >> > The part I love the most about WebKitGTK is that it runs even on a >> > Raspberry Pi 0 but I don't want to build for every platform that has >> been >> > been built by ArchLinux already, I'm not as good as them at configuring >> > build targets. >> >> :-) >> >> > Thanks for any hint, still I'd like to know if ES2015 modules are at >> least >> > in the pipeline, if possible. >> >> I look very quickly into this (just a quick grepping of the ChangeLogs). >> It >> used to be a build-time option, and at some point suppot for ES6 modules >> was >> turned into a runtime option, as an outcome of this bug: >> >> https://bugs.webkit.org/show_bug.cgi?id=164827 >> >> then later on, the runtime setting was removed and support for modules got >> *always* built and enabled. I would expect modules to be working, or to be >> very easy to get working in case we might be missing some port-specific >> code. >> >> I guess we can try to figure this out at some point, but no ETA for now. >> >> Cheers, >> >> -- >> Adrián >> >> ___ >> webkit-gtk mailing list >> webkit-gtk@lists.webkit.org >> https://lists.webkit.org/mailman/listinfo/webkit-gtk >> >> > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK+ VS Apple WebKit
I can confirm it's actually the most advanced browser out there. http://kangax.github.io/compat-table/es6/ 100% score in ES2015 100% score in ES2016+ 100% score in Intl I will bring this under HTML5Test attention. Thanks everyone for info and hints. Best Regards On Wed, Apr 5, 2017 at 11:45 AM, Andrea Giammarchi < andrea.giammar...@gmail.com> wrote: > That'd be awesome news. I'll double check later on and eventually confirm > it (if needed). > > Best Regards > > On Wed, Apr 5, 2017 at 11:02 AM, Yusuke SUZUKI <utatane@gmail.com> > wrote: > >> I believe ES6 modules are enabled in all the platforms. >> >> On Wed, Apr 5, 2017 at 6:08 PM, Adrian Perez de Castro <ape...@igalia.com >> > wrote: >> >>> On Tue, 4 Apr 2017 23:47:29 +0100, Andrea Giammarchi < >>> andrea.giammar...@gmail.com> wrote: >>> >>> > The part I love the most about WebKitGTK is that it runs even on a >>> > Raspberry Pi 0 but I don't want to build for every platform that has >>> been >>> > been built by ArchLinux already, I'm not as good as them at configuring >>> > build targets. >>> >>> :-) >>> >>> > Thanks for any hint, still I'd like to know if ES2015 modules are at >>> least >>> > in the pipeline, if possible. >>> >>> I look very quickly into this (just a quick grepping of the ChangeLogs). >>> It >>> used to be a build-time option, and at some point suppot for ES6 modules >>> was >>> turned into a runtime option, as an outcome of this bug: >>> >>> https://bugs.webkit.org/show_bug.cgi?id=164827 >>> >>> then later on, the runtime setting was removed and support for modules >>> got >>> *always* built and enabled. I would expect modules to be working, or to >>> be >>> very easy to get working in case we might be missing some port-specific >>> code. >>> >>> I guess we can try to figure this out at some point, but no ETA for now. >>> >>> Cheers, >>> >>> -- >>> Adrián >>> >>> ___ >>> webkit-gtk mailing list >>> webkit-gtk@lists.webkit.org >>> https://lists.webkit.org/mailman/listinfo/webkit-gtk >>> >>> >> > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
[webkit-gtk] Fwd: Web with WebGL issues
Running GNOME Web 3.24.3 on ArchLinux, and it's based on WebKitGTK+ 2.16.6 The most basic ever Web page to test WebGL functionality is super flicky and broken: https://get.webgl.org/ I am on Mesa 17.1.5 and Intel hardware, GNOME is running on Wayland instead of Xorg. Dell XPS 13 (2nd gen, core i7) Apologies if already known as issue. Best Regards ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
[webkit-gtk] Dark Theme and Web input
I find it hilarious using Dark Theme means nobody can see what they write on any form online. The usage of `input { color: ButtonText; }` also makes absolute no sense to any Web standard, but it's the only way non styled input can show their text on any regular webpage. Create a page like the following on GNOME tweaked with adwaita dark theme and have fun reading your own input: ```html ``` Is there any plan to make WebKit smart enough to have native dark theme inputs that are readable by default? To be honest, this is one of the main reason I still use Chromium on GNOME instead of Web, which also has superior ECMAScript support and native scripts modules. Thanks in advance for any sort of outcome. Best Regards ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] Dark Theme and Web input
> Do you have any suggestion for what specifically we should do? I guess setting up a top level `input { color: ButtonText; }` style any page can overwrite through its own content has been explored already, right ? I can inject JS upfront too via WebKitGTK+ before any page is loaded, maybe that could be an entry point to pre-style web pages? something like the following: ```js // GJS example based on imports.gi.WebKit2 webView.connect('load-changed', (webView, loadEvent, data) => { switch (loadEvent) { case WebKit2.LoadEvent.COMMITTED: webView.run_javascript( `(function(document){'use strict'; document.addEventListener( 'DOMContentLoaded', e => { const html = document.documentElement; const style = document.createElement('style'); style.textContent = 'input { color: ButtonText; }'; html.insertBefore(style, html.firstChild); }, {once: true} ); }(this.document));`, null, (webView, result) => { webView.run_javascript_finish(result); } ); break; } }); ``` Will have a look at the C++ source too, thanks On Sun, Jul 30, 2017 at 8:56 PM, Michael Catanzaro <mcatanz...@igalia.com> wrote: > On Sun, Jul 30, 2017 at 8:14 PM, Andrea Giammarchi < > andrea.giammar...@gmail.com> wrote: > > Is there any plan to make WebKit smart enough to have native dark theme > inputs that are readable by default? > > > Nope. We all know this is broken, but don't know what to change. Do you > have any suggestion for what specifically we should do? > > The relevant code is mostly in Source/WebCore/rendering/ > RenderThemeGtk.cpp. > > Michael > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] GTK4
Out of curiosity, does that mean GTK4 doesn't bring interesting benefits so nobody cares about having a browser there? On Fri, Nov 23, 2018 at 6:09 PM Michael Catanzaro wrote: > On Thu, Nov 22, 2018 at 12:12 PM, Eric Williams > wrote: > > I was wondering -- has work on a GTK4 port of WebKitGTK started? If > > not, what is the timeline for such work? > > Hi Eric, > > There is no work yet on this and it is not scheduled. I expect it would > be a lot of effort to get working, and we're fully booked right now, > and unfortunately none of our customers care about GTK+ 4 (it's a shame > that Red Hat isn't a customer :) so it seems unlikely that we'd be able > to start on this within the next year. Maybe. > > If you want to help, I know Adrian Perez and Benjamin Otte have been > throwing around some ideas. The plan involved writing a GTK+ 4 based > libwpe backend so that the GTK port would become based on WPE. Adrian, > this thread would be a good chance to document your ideas. > > Michael > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
[webkit-gtk] Web Inspector broken in various distributions
I couldn't find anything online related to this, and apologies if this is not the best place to discuss this, but I think Epiphany Inspector has been broken for a very long time in latest GNOME, in both ArchLinux and/or Ubuntu. Is there some dependency I am missing, in order to have the ability to use the Web inspector? The first time I click on "Inspect Element" nothing happens, the second time I have this at the bottom of any page I try to inspect: This page contains the following errors: error on line 43 at column 8: Opening and ending tag mismatch: link line 0 and head Below is a rendering of the page up to the first error. WI.sharedApp = new WI.AppController; WI.sharedApp.initialize(); Thanks in advance for any sort of hint, suggestion, solution, or indication. Best Regards ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] Web Inspector broken in various distributions
Thanks a lot, it looks like all I have to do is waiting for changes to land in ArchLinux. Best Regards On Tue, Oct 29, 2019 at 12:23 PM Carlos Garcia Campos wrote: > El mar, 29-10-2019 a las 11:58 +0100, Andrea Giammarchi escribió: > > I couldn't find anything online related to this, and apologies if > > this is not the best place to discuss this, but I think Epiphany > > Inspector has been broken for a very long time in latest GNOME, in > > both ArchLinux and/or Ubuntu. > > > > Is there some dependency I am missing, in order to have the ability > > to use the Web inspector? > > It's a shared-mime-info bug, see > > https://bugs.webkit.org/show_bug.cgi?id=202321 > > > The first time I click on "Inspect Element" nothing happens, the > > second time I have this at the bottom of any page I try to inspect: > > > > This page contains the following errors: > > error on line 43 at column 8: Opening and ending tag mismatch: link > > line 0 and head > > Below is a rendering of the page up to the first error. > > WI.sharedApp = new WI.AppController; WI.sharedApp.initialize(); > > > > > > Thanks in advance for any sort of hint, suggestion, solution, or > > indication. > > > > Best Regards > > ___ > > webkit-gtk mailing list > > webkit-gtk@lists.webkit.org > > https://lists.webkit.org/mailman/listinfo/webkit-gtk > -- > Carlos Garcia Campos > http://pgp.rediris.es:11371/pks/lookup?op=get=0xF3D322D0EC4582C3 > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WebKitGTK 2.26.4 released!
> Always use a light theme for rendering form controls. eventually congrats for the release On Fri, Feb 14, 2020 at 11:11 AM Carlos Garcia Campos wrote: > WebKitGTK 2.26.4 is available for download at: > > https://webkitgtk.org/releases/webkitgtk-2.26.4.tar.xz (18.4MB) >md5sum: 60f881729f3b71244b7f6e58790073e0 >sha1sum: 72f209c08ecc8ad4f0f6b767d4fa1be7a652df33 >sha256sum: > 4386900713dfadf9741177210b32623cab22562a79ffd0d446b66569934b113f > > This is a bug fix release in the stable 2.26 series. > > What's new in the WebKitGTK 2.26.4 release? > === > > - Always use a light theme for rendering form controls. > - Fix the build with WPE renderer disabled. > - Fix the build with OpenGL disabled. > - Fix the build with GCC 10. > - Fix several crashes and rendering issues. > > What is WebKitGTK? > == > > WebKitGTK is the GNOME platform port of the WebKit rendering engine. > Offering WebKit's full functionality through a set of GObject-based > APIs, it is suitable for projects requiring any kind of web > integration, from hybrid HTML/CSS applications to full-fledged web > browsers. > > More information > > > If you want to know more about the project or get in touch with us > you may: > > - Visit our website at https://www.webkitgtk.org, or the upstream > site at https://www.webkit.org. People interested in contributing > should read: https://www.webkit.org/coding/contributing.html. > > - Browse the bug list at https://bugs.webkit.org, WebKitGTK bugs are > typically prefixed by "[GTK]." A bug report with a minimal, > reproducible test case is often just as valuable as a patch. > > - Join the #webkit and #webkitgtk IRC channels at irc.freenode.net. > > - Subscribe to the WebKitGTK mailing list, > https://lists.webkit.org/mailman/listinfo/webkit-gtk, or the > WebKit development mailing list, > https://lists.webkit.org/mailman/listinfo/webkit-dev > > Thanks > == > > Thanks to all the contributors who made possible this release, they > are far too many to list! > > The WebKitGTK team, > February 14, 2020 > -- > Carlos Garcia Campos > http://pgp.rediris.es:11371/pks/lookup?op=get=0xF3D322D0EC4582C3 > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] Hardware acceleration on raspberry pi 4 using buildroot.
Oh wow, I didn't know that myself ... this might be a game-changer for me as well, as it sadden me I had no WPE option via GJS, but apparently there is one, so I'll try to build it too. If you're considering ArchLinux, I might put such build in AUR, as they removed wpebwekit, completely destroying my automated ability to build the bin, which I've done for a year or more. https://aur.archlinux.org/packages/wpewebkit-bin/ Thanks for coming back with this info! On Mon, Jun 7, 2021 at 2:09 PM C Larbi wrote: > Thanks for your suggestions Andrea. > I followed up on your WPE comment and found out that there is a build > switch called "-DUSE_WPE_RENDERER" to let webkitgtk use the WPE renderer. > > In my buildroot configuration file, this switch is off, but further > reading revealed that upstream recommends the use of this option. > > I am away from my build system currently, but will give it a try (turn on > wpe renderer switch) when i get back > > Regards > Cerezo > > On Mon, Jun 7, 2021, 8:44 AM Andrea Giammarchi < > andrea.giammar...@gmail.com> wrote: > >> Not sure if relevant, or a proper answer, 'cause I think there are no >> bindings for Gtk (or maybe there are?) but the fastest browser for the Pi >> (2, 3, 4) is WPEWebKit, which is HW accelerated. >> To date, I've never managed to get nearly similar performance via >> WebKit2-GTK: not CSS, not Canvas, even less WebGL, video are slow. >> >> Maybe you can bootstrap COG and WPE a part? For all those cases I've >> mentioned, it does an excellent job (30fps+ stable on full screen WebGL, >> and excellent video playback + smooth CSS animations). >> >> On Mon, Jun 7, 2021 at 1:08 AM C Larbi wrote: >> >>> I am working on an information display screen using webkitgtk. >>> >>> The system was put together using buildroot. >>> >>> The display screen uses some javacript to manipulate the DOM to create >>> some effects. >>> >>> So far everthing seems to be working perfectly except that it seems >>> rendering the effects seem to be slow. >>> >>> Some of these effects are to give smooth transitions, but on my pi 4, >>> these transitions are seen as visible STEPS instead of being smooth. >>> >>> I am no expert in GPUs but it seems the gpu is not being utilized in >>> rendering. >>> >>> When i set the environment variable "WEBKIT_FORCE_COMPOSITING_MODE" to >>> 1, the whole screen gets garbled. >>> >>> I will appreciate any pointers in getting this fixed >>> >>> Below are some details of my system: >>> >>> webkitgtk version = 2.30.2 >>> Mesa3d version = 20.2.2 >>> Opengl es and opengl egl are enabled >>> Using wayland compositor (weston) >>> Using the vc4-fkms-v3d on pi 4 >>> >>> Regards >>> Cerezo >>> >>> ___ >>> webkit-gtk mailing list >>> webkit-gtk@lists.webkit.org >>> https://lists.webkit.org/mailman/listinfo/webkit-gtk >>> >> ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] Hardware acceleration on raspberry pi 4 using buildroot.
Not sure if relevant, or a proper answer, 'cause I think there are no bindings for Gtk (or maybe there are?) but the fastest browser for the Pi (2, 3, 4) is WPEWebKit, which is HW accelerated. To date, I've never managed to get nearly similar performance via WebKit2-GTK: not CSS, not Canvas, even less WebGL, video are slow. Maybe you can bootstrap COG and WPE a part? For all those cases I've mentioned, it does an excellent job (30fps+ stable on full screen WebGL, and excellent video playback + smooth CSS animations). On Mon, Jun 7, 2021 at 1:08 AM C Larbi wrote: > I am working on an information display screen using webkitgtk. > > The system was put together using buildroot. > > The display screen uses some javacript to manipulate the DOM to create > some effects. > > So far everthing seems to be working perfectly except that it seems > rendering the effects seem to be slow. > > Some of these effects are to give smooth transitions, but on my pi 4, > these transitions are seen as visible STEPS instead of being smooth. > > I am no expert in GPUs but it seems the gpu is not being utilized in > rendering. > > When i set the environment variable "WEBKIT_FORCE_COMPOSITING_MODE" to 1, > the whole screen gets garbled. > > I will appreciate any pointers in getting this fixed > > Below are some details of my system: > > webkitgtk version = 2.30.2 > Mesa3d version = 20.2.2 > Opengl es and opengl egl are enabled > Using wayland compositor (weston) > Using the vc4-fkms-v3d on pi 4 > > Regards > Cerezo > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WPEWebKit goes half speed
Dear all, I have finally solved the problem, or better, Zan found what was causing the performance drop, and turned out ICU was constantly querying the FS in search for /etc/localtime which, if not present, was causing a lot of back and forth, degrading performance. The solution, which has been updated in my ArchLinux kiosk installer for the Pi, was to use timedatectl [1] and be sure such a symlink was there. sudo timedatectl set-timezone "$(curl --fail https://ipapi.co/timezone)" Thanks to Zan in particular, and thanks to this ML too. Have a lovely weekend [1] https://wiki.archlinux.org/title/System_time#Update_timezone_every_time_NetworkManager_connects_to_a_network On Mon, Aug 9, 2021 at 3:48 PM wrote: > On Sun, Aug 8, 2021, at 11:11 AM, Andrea Giammarchi wrote: > > cog 0.6 and 2.28 (IIRC) were just fine ... the raspbian image from the > wpewebkit site, which has cog 0.6, goes, on the same hardware, more than > twice as fast. > > Meanwhile, weston 9 happened, mesa brought GLES 3.1 in the game, so it's > hard to tell, in isolation, what is causing such drop, but I guess there's > nothing bad in the way I build WPEWebKit there, right? > > Thanks for the reply though. > > > No, you build it just fine. > > Removal of ACCELERATED_2D_CANVAS option shouldn't have any effect on WebGL > content. It was only ever used for 2D canvas, and even there it had varying > performance with stability/quality problems. I don't know what the state of > it was in 2.28 but when I was removing the feature it wasn't really working > anyway. > > On Wed, Aug 4, 2021 at 9:58 PM Fujii Hironori > wrote: > > > On Tue, Aug 3, 2021 at 6:57 PM Andrea Giammarchi < > andrea.giammar...@gmail.com> wrote: > > Dear list, > I've recently built wpewebkit 2.32.3 [1] and cog 0.10 [2] and it > starts/runs fine on weston in both Pi 4 32bit, Pi4 64bit, and Pi3 64bit. > > However, the jellyfish demo [3] that once run at ~60fps [4] is now ~24FPS > on Pi 4, and ~15FPS on Pi 3. I am not using dedicated cairo glesv2 builds, > just regular cairo, latest mesa drivers, vulkan, and so on. > > > Which version was previously working fine? > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
[webkit-gtk] WPEWebKit goes half speed
Dear list, I've recently built wpewebkit 2.32.3 [1] and cog 0.10 [2] and it starts/runs fine on weston in both Pi 4 32bit, Pi4 64bit, and Pi3 64bit. However, the jellyfish demo [3] that once run at ~60fps [4] is now ~24FPS on Pi 4, and ~15FPS on Pi 3. I am not using dedicated cairo glesv2 builds, just regular cairo, latest mesa drivers, vulkan, and so on. Does anyone know how come latest builds produces such drop? Only changes I've seen is that ENABLE_2D_CANVAS_ACCELRATION is gone/ignored, as parameter, but while everything works OK and seems to be HW accelerated (compared to regular WebKitGTK that goes ~9fps), I cannot run smoothly WebGL anymore (haven't tested videos, but these also were working fine at 720p in all platforms). Thanks in advance for any kind of hint/direction and, in case it helps, this is how I build WPEWebKit in all the platforms [5]. Regards, Andrea. [1] https://aur.archlinux.org/packages/wpewebkit-bin/ [2] https://aur.archlinux.org/packages/cog-wpe-bin [3] https://archibold.io/demo/jellyfish/ [4] https://youtu.be/T3BdiEe8lqA?t=30 [5] https://github.com/WebReflection/archibold.io/blob/master/aur/build-wpewebkit-bin#L81-L91 ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] WPEWebKit goes half speed
cog 0.6 and 2.28 (IIRC) were just fine ... the raspbian image from the wpewebkit site, which has cog 0.6, goes, on the same hardware, more than twice as fast. Meanwhile, weston 9 happened, mesa brought GLES 3.1 in the game, so it's hard to tell, in isolation, what is causing such drop, but I guess there's nothing bad in the way I build WPEWebKit there, right? Thanks for the reply though. On Wed, Aug 4, 2021 at 9:58 PM Fujii Hironori wrote: > > On Tue, Aug 3, 2021 at 6:57 PM Andrea Giammarchi < > andrea.giammar...@gmail.com> wrote: > >> Dear list, >> I've recently built wpewebkit 2.32.3 [1] and cog 0.10 [2] and it >> starts/runs fine on weston in both Pi 4 32bit, Pi4 64bit, and Pi3 64bit. >> >> However, the jellyfish demo [3] that once run at ~60fps [4] is now ~24FPS >> on Pi 4, and ~15FPS on Pi 3. I am not using dedicated cairo glesv2 builds, >> just regular cairo, latest mesa drivers, vulkan, and so on. >> >> > Which version was previously working fine? > > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
[webkit-gtk] WPEWebKit 2.35.1 install issue
After a successful build from scratch, it fails at installing. The building commands are the same since ages ago, and I've only enabled AVIF and JPEGXL. https://github.com/WebReflection/archibold.io/blob/master/aur/build-wpewebkit-bin#L82-L108 And this it the end of the building part: [3/4] Install the project... -- Install configuration: "Release" CMake Error at Source/cmake_install.cmake:46 (file): file INSTALL cannot find "/home/ag/aur/wpewebkit/build/Documentation/wpe-1.1/html/wpe-1.1": No such file or directory. Call Stack (most recent call first): cmake_install.cmake:47 (include) FAILED: CMakeFiles/install.util Thanks in advance for kind of help/hint ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] No flag for WebGPU - is it planned?
Thanks, although I see `_Metal` or `_METAL` is a bit all over the repository ... and declared as backend for the WebGPU folder: https://github.com/WebKit/WebKit/blob/698b77b55d9bc0ab42b6f6163de772de2512793f/Source/WebGPU/WebGPU/Adapter.mm#L77 May I ask if anyone is even thinking about enabling that Vulkan one and/or bringing that WebGPU in? https://github.com/WebKit/WebKit/blob/698b77b55d9bc0ab42b6f6163de772de2512793f/Source/WebGPU/WebGPU/WebGPU.h#L192C16-L203 Would a custom build of WebKitGTK that replace Metal with Vulkan adapter work? I understand I might not have answers around these topics, but as there is a folder and clearly right now everything floats around Metal API only in a static way, I am more than curious about why all those options or if these would anyhow works once exposed. Today I had to use Chromium Dev channel on Linux with Vulkan and unsafe WebGPU flags enabled to run ML out of my browser to test Web Stable Diffusion https://websd.mlc.ai/ and I was super sad about not being able to do so with my daily driver: GNOME Web. Thanks. On Fri, Oct 6, 2023 at 2:00 PM Michael Catanzaro wrote: > > Hi Andrea, > > WebGPU is not supported and nobody is working on it, so I assume it's > also not planned. Sorry. > > Michael > > > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
[webkit-gtk] No flag for WebGPU - is it planned?
Looking at all flags in here https://github.com/WebKit/WebKit/blob/main/Source/JavaScriptCore/runtime/OptionsList.h I couldn't find anything mentioning WebGPU which is instead enabled in Safari. I wonder if anyone could gently summarize the current status or ETA for having such flag available, unless it's something enabled by default behind the scene. I have also seen a WebGPU folder so I think there is something but I wouldn't know how to test or enable such something (I believe based on Vulkan). Thanks in advance for any sort of pointer, info, or clarification. ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] Enabling SharedArrayBufer in GNOME Web (epiphany)
AWESOME, thank you! P.S. I believe this should be a flag in preferences like it is in Safari but at least now I know what to do here, thanks again. On Tue, Jun 20, 2023 at 12:33 PM Philippe Normand wrote: > On Tue, 2023-06-20 at 11:14 +0200, Andrea Giammarchi wrote: > > I have tried both Epiphany Technology Preview 44.0 with WebKitGTK > > 2.41.5 and stable Web 44.3+ with WebKitGTK 2.40.2 and nowhere in > > preferences, neither in about:memory, nor in CLI, I could find a way > > to enable SharedArrayBuffer like people using Safari can/do. > > > > I am developing JS/WASM ffi / interpreters / bindings and I am > > incapable of using Web as daily driver or even test whatever I write > > is actually working in there, as the SharedArrayBuffer is mandatory > > to enable Atomics and Worker communication. > > > > Can anyone please help? > > export JSC_useSharedArrayBuffer=1 (or true?)? > See Source/JavaScriptCore/runtime/OptionsList.h > > > > > Thank you! > > ___ > > webkit-gtk mailing list > > webkit-gtk@lists.webkit.org > > https://lists.webkit.org/mailman/listinfo/webkit-gtk > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
[webkit-gtk] Enabling SharedArrayBufer in GNOME Web (epiphany)
I have tried both Epiphany Technology Preview 44.0 with WebKitGTK 2.41.5 and stable Web 44.3+ with WebKitGTK 2.40.2 and nowhere in preferences, neither in about:memory, nor in CLI, I could find a way to enable SharedArrayBuffer like people using Safari can/do. I am developing JS/WASM ffi / interpreters / bindings and I am incapable of using Web as daily driver or even test whatever I write is actually working in there, as the SharedArrayBuffer is mandatory to enable Atomics and Worker communication. Can anyone please help? Thank you! ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] webkit-gtk for Windows
FWIWI I did succeed at some point via WSL and Xming ... if interested: https://webreflection.medium.com/testing-safari-via-bash-for-windows-js-2b7b9098d9c0 On Mon, Feb 5, 2024 at 12:25 AM Michael Catanzaro wrote: > > Hi, > > The answer is no. > > You want to use the Windows port instead: > https://docs.webkit.org/Ports/WindowsPort.html > > Michael > > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk
Re: [webkit-gtk] Displaying very long HTML documents (books), etc.
Answering from memories (TweetDeck infinite scroll challenge) the biggest challenge is not to incrementally populate the view with previous/next chapter in the book/stream, the issue is how are you going to handle that scrollbar. There's no hint on browsers to tell you that the handler reached its minimum size (AFAIK and IIRC), and if you base your computation on an empty document filled with new lines to match that length it's very likely you'll have issues when words wrapping would instead happen in the document so that you can't really compute that difference between the scroll bar position and the visualized text. More issues come from the Ctrl+F (search) builtin implementation, so that what seems to be a "just render only what's visible on the current area" is a way bigger problem to deal with ... infinite scrolling pages have the luck to be able to re-map each chunk of the stream as well-known length, but a whole book that is already known in advance but not 100% rendered yet, would be challenging to map to a reliable scrollbar. Apologies if I didn't help at all, but the gist of my comment is: "indeed try to see if you have a problem before having dozen more to deal with". On Wed, Feb 7, 2024 at 3:18 PM Michael Catanzaro wrote: > On Wed, Feb 7 2024 at 03:54:40 PM +02:00:00, Sim Tov > wrote: > > 1. What is the maximal HTML document size that still can be displayed > > smoothly? > > 2. Is there a hard limit enforced by WebkitGT > > There is no max size. Loading performance will get worse as the max > document size increases. My suggestion is to just try it and see if > it's a problem in practice or not. My suspicion is that it will be able > to handle even extremely long books adequately, but I don't know for > sure. > > > 3. Are you aware of any apps implementing the two features mentioned > > in the previous email (so I can look for examples/ideas there)? > > No, sorry. > > Michael > > > ___ > webkit-gtk mailing list > webkit-gtk@lists.webkit.org > https://lists.webkit.org/mailman/listinfo/webkit-gtk > ___ webkit-gtk mailing list webkit-gtk@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-gtk