User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Firefox/78.0

Steps to reproduce:

I downloaded the latest source code from nighlty branch and compiled it, on linux-amd64.

Actual results:

I got this error:

24:59.75 In file included from /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/nsISupportsUtils.h:17,
24:59.75 from /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/nsISupports.h:82,
24:59.75 from /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/CompositorWidget.h:8,
24:59.75 from /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsWindow.h:14,
24:59.76 from /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsWindow.cpp:8:
24:59.76 /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/mozilla/RefPtr.h: In instantiation of 'static void mozilla::RefPtrTraits<U>::AddRef(U*) [with U = DataOffer]':
24:59.76 /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/mozilla/RefPtr.h:380:67: required from 'static void RefPtr<T>::ConstRemovingRefPtrTraits<U>::AddRef(U*) [with U = DataOffer; T = DataOffer]'
24:59.77 /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/mozilla/RefPtr.h:97:43: required from 'RefPtr<T>::RefPtr(const RefPtr<T>&) [with T = DataOffer]'
24:59.77 /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsWindow.cpp:7916:67: required from here
24:59.77 /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/mozilla/RefPtr.h:49:39: error: invalid use of incomplete type 'class DataOffer'
24:59.77 49 | static void AddRef(U* aPtr) { aPtr->AddRef(); }
24:59.77 | ^
24:59.77 In file included from /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsWindow.cpp:66:
24:59.77 /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsDragService.h:17:7: note: forward declaration of 'class DataOffer'
24:59.77 17 | class DataOffer;
24:59.78 | ^~~~~~~~~
24:59.78 In file included from /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/nsISupportsUtils.h:17,
24:59.78 from /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/nsISupports.h:82,
24:59.78 from /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/CompositorWidget.h:8,
24:59.78 from /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsWindow.h:14,
24:59.78 from /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsWindow.cpp:8:
24:59.79 /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/mozilla/RefPtr.h: In instantiation of 'static void mozilla::RefPtrTraits<U>::Release(U*) [with U = DataOffer]':
24:59.79 /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/mozilla/RefPtr.h:381:69: required from 'static void RefPtr<T>::ConstRemovingRefPtrTraits<U>::Release(U*) [with U = DataOffer; T = DataOffer]'
24:59.79 /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/mozilla/RefPtr.h:81:44: required from 'RefPtr<T>::RefPtr() [with T = DataOffer]'
24:59.79 /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsWindow.cpp:7916:67: required from here
24:59.80 /var/tmp/portage/www-client/firefox-92.0/work/firefox_build/dist/include/mozilla/RefPtr.h:50:40: error: invalid use of incomplete type 'class DataOffer'
24:59.80 50 | static void Release(U* aPtr) { aPtr->Release(); }
24:59.80 | ^
24:59.80 In file included from /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsWindow.cpp:66:
24:59.80 /var/tmp/portage/www-client/firefox-92.0/work/firefox-92.0/widget/gtk/nsDragService.h:17:7: note: forward declaration of 'class DataOffer'
24:59.80 17 | class DataOffer;
24:59.80 | ^

Expected results:

I expect the compile to succed. The compressed build log is attached.

I'm uncertain where this may have been introduced, but widgets/gtk/ is activly developed at the moment! :)

tried again with clang-12.0.1, same error I believe. But maybe it has better error message?

The Bugbug bot thinks this bug should belong to the 'Core::Widget: Gtk' component, and is moving the bug to that component. Please revert this change in case you think the bot is wrong.

same build failure on openbsd, cf http://buildbot.rhaalovely.net/nine/#/builders/3/builds/1152

I suppose these are X11 only builds (no Wayland)?

that is true for me at least, I'm using --enable-default-toolkit=cairo-gtk3

There is more X11-only bustage:

In file included from Unified_cpp_widget_gtk0.cpp:83:
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:234:16: error: use of undeclared identifier 'WL_OUTPUT_MODE_CURRENT'
  if ((flags & WL_OUTPUT_MODE_CURRENT) == 0) return;
               ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:240:40: error: variable has incomplete type 'const struct wl_output_listener'
static const struct wl_output_listener output_listener = {
                                       ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:240:21: note: forward declaration of 'mozilla::widget::wl_output_listener'
static const struct wl_output_listener output_listener = {
                    ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:247:49: error: unknown type name 'wl_registry'
static void screen_registry_handler(void* data, wl_registry* registry,
                                                ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:253:55: error: use of undeclared identifier 'wl_output_interface'
        WaylandRegistryBind<wl_output>(registry, id, &wl_output_interface, 2);
                                                      ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:268:42: error: variable has incomplete type 'const struct wl_registry_listener'
static const struct wl_registry_listener screen_registry_listener = {
                                         ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:268:21: note: forward declaration of 'mozilla::widget::wl_registry_listener'
static const struct wl_registry_listener screen_registry_listener = {
                    ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:273:3: error: unknown type name 'wl_display'
  wl_display* display = WaylandDisplayGetWLDisplay();
  ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:273:25: error: use of undeclared identifier 'WaylandDisplayGetWLDisplay'
  wl_display* display = WaylandDisplayGetWLDisplay();
                        ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:299:31: error: unknown type name 'wl_registry_destroy'; did you mean 'wl_registry_listener'?
  g_clear_pointer(&mRegistry, wl_registry_destroy);
                              ^~~~~~~~~~~~~~~~~~~
                              wl_registry_listener
/home/glandium/.mozbuild/sysroot-x86_64-linux-gnu/usr/include/glib-2.0/glib/gmem.h:118:49: note: expanded from macro 'g_clear_pointer'
    GDestroyNotify _destroy = (GDestroyNotify) (destroy);                      \
                                                ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:268:21: note: 'wl_registry_listener' declared here
static const struct wl_registry_listener screen_registry_listener = {
                    ^
/tmp/central/widget/gtk/ScreenHelperGTK.cpp:299:3: error: expected expression
  g_clear_pointer(&mRegistry, wl_registry_destroy);
  ^
/home/glandium/.mozbuild/sysroot-x86_64-linux-gnu/usr/include/glib-2.0/glib/gmem.h:118:57: note: expanded from macro 'g_clear_pointer'
    GDestroyNotify _destroy = (GDestroyNotify) (destroy);                      \
                                                        ^
9 errors generated.

If you want to fool around with a X11-only gtk build on try, use the following patch:
https://hg.mozilla.org/try/rev/35650b3991fc07561946bfd6a79dac8f86c0597d

so does that mean --enable-default-toolkit=cairo-gtk3 is outlawed now?

(In reply to tt_1 from comment #8)

so does that mean --enable-default-toolkit=cairo-gtk3 is outlawed now?

dont think so, it's just that mozilla automation doesnt test anymore X11-only builds, and those failures are prone to reappear here and there during wayland work.

Guys, do you mind to attach a patch for it?
Thanks.

ScreenHelperGTK.cpp is trivially fixable, the DataOffer part not so much...

(In reply to Martin Stránský [:stransky] (ni? me) from comment #10)

Guys, do you mind to attach a patch for it?
Thanks.

what happened to 'you break it, you fix it' ? :D unless wayland-enabled builds are mandatory now ofc ...

(In reply to Landry Breuil (:gaston) from comment #12)

(In reply to Martin Stránský [:stransky] (ni? me) from comment #10)

Guys, do you mind to attach a patch for it?
Thanks.

what happened to 'you break it, you fix it' ? :D unless wayland-enabled builds are mandatory now ofc ...

Sure, it's on my queue.

This is probably just temporary fix to allow X11 build again. Instead of
ifdeds some better abstraction might be needed. Hope that Martin can
find better way.

Thanks for the patch Petr, I tried a different approach.

Comment on attachment 9234120 [details]
Bug 1721707 [Wayland] Build fix for X11 only environment r?PetrSumbera

m-c built fine for me with this patch on OpenBSD, thanks !

Should we finally add a X11-only CI build? :) (and in the future a Wayland-only one, too?)

https://hg.mozilla.org/mozilla-central/rev/fa29a9038d42

(In reply to Robert Mader [:rmader] from comment #20)

Should we finally add a X11-only CI build? :) (and in the future a Wayland-only one, too?)

if its possible and helps not to run into this kind of problem again, yes absolutly please.

(In reply to Robert Mader [:rmader] from comment #20)

Should we finally add a X11-only CI build? :) (and in the future a Wayland-only one, too?)

Filed bug 1723869.