Per Žan's request on the webkit-gtk mailing list, I'm creating this umbrella to track webkit-gtk issues affecting MacPorts' webkit-gtk port on OS X which covers WebKit GTK/X11/darwin and WebKit GTK/Quartz/darwin.
Can this apply to building WebKit GTK/Quartz/Darwin through JHBuild, which is Gnome's official way of building their stack on OSX? If so, then I'd also like to introduce dependencies on the following bugs: https://bugs.webkit.org/show_bug.cgi?id=98262 https://bugs.webkit.org/show_bug.cgi?id=133293 https://bugs.webkit.org/show_bug.cgi?id=138332 https://bugs.webkit.org/show_bug.cgi?id=140167 https://bugs.webkit.org/show_bug.cgi?id=140171 https://bugs.webkit.org/show_bug.cgi?id=140598 I'll be opening more of these as soon as I can produce a working MiniBrowser on OSX with WebKit2.
Here is a comprehensive list of the bugs that currently block building WebKit GTK/Quartz/Darwin for me: (tested on 2.9.1) https://bugs.webkit.org/show_bug.cgi?id=127059 https://bugs.webkit.org/show_bug.cgi?id=138332 https://bugs.webkit.org/show_bug.cgi?id=144554 https://bugs.webkit.org/show_bug.cgi?id=144555 https://bugs.webkit.org/show_bug.cgi?id=144557 https://bugs.webkit.org/show_bug.cgi?id=144560 https://bugs.webkit.org/show_bug.cgi?id=144561 https://bugs.webkit.org/show_bug.cgi?id=144785 https://bugs.webkit.org/show_bug.cgi?id=144846 https://bugs.webkit.org/show_bug.cgi?id=144845 For 2.8.0, the last two bugs in the above list do not apply, but the following one does: https://bugs.webkit.org/show_bug.cgi?id=144559
It would be good to request commit-queue as well as review on the patches that are intended for master, since you're not a committer. I've added it to those that have been reviewed, since I assume that's what you want.
Yes, please. Thanks for the reminder. I'll make sure it's done on the other ones.
Would it be too much asking for a WebKit branch on, say, Github? I'd like to test this and maybe help with the code reviews a bit.
(In reply to comment #5) > Would it be too much asking for a WebKit branch on, say, Github? I'd like to > test this and maybe help with the code reviews a bit. I was trying to ask kindly, obviously. Sorry, I'm not a native English speaker :)
Here's a branch on Github: https://github.com/ptomato/webkit/pull/1 Yes please, any help is welcome. These patches were last known to work on top of 2.9.1 (plus some other ones that have been committed to master in the meantime.) I've rebased them onto current master, but I don't know if it will still work. I'll try building it over the weekend.
Now updated at https://github.com/ptomato/webkit/pull/2
(In reply to comment #8) > Now updated at https://github.com/ptomato/webkit/pull/2 Thanks! I got to the point where the jhbuild is set-up, I had to isolate some modules in a separate linux-only modulates though. Also installed quite a few deps using homebrew. I guess I'll prepare some patches about this soon :) build-webkit now fails here at the bmalloc ar step, because Xcode's ar doesn't have response files support it seems. Does this ring a bell to you? Are you building with Xcode's tools and clang? I'm using Ninja btw. Are you too?
(In reply to comment #9) > (In reply to comment #8) > > Now updated at https://github.com/ptomato/webkit/pull/2 > > Thanks! > > I got to the point where the jhbuild is set-up, I had to isolate some > modules in a separate linux-only modulates though. Also installed quite a > few deps using homebrew. I guess I'll prepare some patches about this soon :) > > build-webkit now fails here at the bmalloc ar step, because Xcode's ar > doesn't have response files support it seems. Does this ring a bell to you? > Are you building with Xcode's tools and clang? I do use the XCode toolchain but as for jhbuild, I use the standard gtk-osx jhbuild modulesets; described here [1]. I haven't needed to use anything installed from homebrew. I have a work-in-progress branch here [2]. (That builds the current stable release 2.8.3, but you can easily change it to build from the all-osx-patches branch on ptomato/webkit.) In particular, this line [3] shows the build settings that I use. > I'm using Ninja btw. Are you too? No, should I? [1] https://wiki.gnome.org/Projects/GTK+/OSX/Building [2] https://github.com/ptomato/gtk-osx-build/tree/update-webkit2 [3] https://github.com/ptomato/gtk-osx-build/blob/update-webkit2/modulesets-stable/gtk-osx-unsupported.modules#L305
(In reply to comment #10) > (In reply to comment #9) > > (In reply to comment #8) > > > Now updated at https://github.com/ptomato/webkit/pull/2 > > > > Thanks! > > > > I got to the point where the jhbuild is set-up, I had to isolate some > > modules in a separate linux-only modulates though. Also installed quite a > > few deps using homebrew. I guess I'll prepare some patches about this soon :) > > > > build-webkit now fails here at the bmalloc ar step, because Xcode's ar > > doesn't have response files support it seems. Does this ring a bell to you? > > Are you building with Xcode's tools and clang? > > I do use the XCode toolchain but as for jhbuild, I use the standard gtk-osx > jhbuild modulesets; described here [1]. I haven't needed to use anything > installed from homebrew. I have a work-in-progress branch here [2]. (That > builds the current stable release 2.8.3, but you can easily change it to > build from the all-osx-patches branch on ptomato/webkit.) > Sure but if we ever want to have an EWS or buildbot or more generally ease contributions from Mac users we need build-webkit working with the internal jhbuild, IMHO. > In particular, this line [3] shows the build settings that I use. > Yes I've seen it. I'll enable video and web audio though :) > > I'm using Ninja btw. Are you too? > > No, should I? > Ha, well, maybe not :) Seems like I'm hitting a bug in the cmake ninja generator. Unininstalling ninja and kicking a new build got me a bit further now.
(In reply to comment #11) > (In reply to comment #10) > > (In reply to comment #9) > > > (In reply to comment #8) > > > > Now updated at https://github.com/ptomato/webkit/pull/2 > > > > > > Thanks! > > > > > > I got to the point where the jhbuild is set-up, I had to isolate some > > > modules in a separate linux-only modulates though. Also installed quite a > > > few deps using homebrew. I guess I'll prepare some patches about this soon :) > > > > > > build-webkit now fails here at the bmalloc ar step, because Xcode's ar > > > doesn't have response files support it seems. Does this ring a bell to you? > > > Are you building with Xcode's tools and clang? > > > > I do use the XCode toolchain but as for jhbuild, I use the standard gtk-osx > > jhbuild modulesets; described here [1]. I haven't needed to use anything > > installed from homebrew. I have a work-in-progress branch here [2]. (That > > builds the current stable release 2.8.3, but you can easily change it to > > build from the all-osx-patches branch on ptomato/webkit.) > > > > Sure but if we ever want to have an EWS or buildbot or more generally ease > contributions from Mac users we need build-webkit working with the internal > jhbuild, IMHO. Yes, maybe you could copy the moduleset from gtk-osx-build?
I'd like to add dependencies on the following bugs, please: https://bugs.webkit.org/show_bug.cgi?id=148606 https://bugs.webkit.org/show_bug.cgi?id=148607
I've had some success finally tweaking a bit the internal JHBuild set-up we have and combining it with Homebrew. I still need to clean that patch up though.
Those following this tracker might like to also know that current versions of webkit-gtk fail to build with current versions of g-ir-scanner with errors like this: ld: unknown option: -rpath=. See https://bugzilla.gnome.org/show_bug.cgi?id=625195 for a patch (also just pushed to MacPorts).
Hello, I have created bug https://bugs.webkit.org/show_bug?id=153093 , I was wondering if it could be added to this tracker please.