svn commit: r490910 - in head: Mk Mk/Uses mail/thunderbird www/firefox www/firefox-esr www/firefox-i18n www/firefox/files www/palemoon www/seamonkey

[Jan Beich]

Martin Wilke <miwi at FreeBSD.org> writes:

> On Tue, Jan 22, 2019 at 09:52:11PM +0100, Christoph Moench-Tegeder wrote:
>
>> ## Jan Beich (jbeich at FreeBSD.org):
>>
>> > Looking at Try options[1] and automation[2] I'm not sure WAYLAND=off is
>> > tested nowadays. In www/firefox WAYLAND=off has no effect[3] unless
>> > x11-toolkits/gtk30 also has WAYLAND=off. I've reported the issue
>> > upstream[4] but for now just locally remove files/patch-bug1514156.
>> > If there's no fix in sight after actual release on 2019-01-29 then
>> > the patch can be moved under WAYLAND_EXTRA_PATCHES.
>>
>> I personally don't mind having the wayland stuff around - from
>> the log I gather I disabled it quite some time ago because it
>> broke something ("glx/dri/whatever", sic) back in the days.
>> Now it just works. And firefox 65 works, too.
>>
>> Given that WAYLAND is the default for us, and "without wayland"
>> might be untested anyways - would it make sense to remove the
>> WAYLAND option and force it to "on" always?
>
> Correct me if I am wrong, WAYLAND is a X replacement which is not really working with FreeBSD atm.

x11-wm/sway works, except on x11/nvidia-driver. No clue about others.

GDK_BACKEND=wayland is default in Gtk3 if running under Wayland, except
www/firefox. QT_QPA_PLATFORM=wayland-egl and SDL_VIDEODRIVER=wayland
often work fine but not default. Lots of X11 apps work under Xwayland
e.g., www/seamonkey, editors/emacs, x11/xterm.

> Why should I/We install it when we dont use it? For me the only thing it does is increasing build time,
> and for minimal desktops a useless dependency. Its already absurd enough to have 2 CLANG versions
> on a desktop.

For minimal X11 desktops a lot of stuff is useless e.g., D-Bus per bug 194460
but accessibility/orca needs D-Bus. I suspect all Wayland-specific dependencies
*combined* don't reach the build time of a single devel/llvm*.