[vlc-devel] [PATCH 1/4] libvlc: Add "fullscreen-monitor" setting

Tue Oct 15 21:02:33 CEST 2019

Hi,

On Tue, Oct 15, 2019 at 10:06:37AM +0300, Rémi Denis-Courmont wrote:
> Hi,
>
> As noted in the last or previous to last video workshop, the whole point of separating open and enable in window provider was to support this - just like why I split open and start in audio outputs earlier.

I'm not sure it was in the previous workshop as I don't have it in my notes, but
I might have missed this part. It's not important though as long as things can
be clear on the mailing list.

I understand that this is like the audio outputs, it's the obvious part. The
rest of the design is more blurred, likely because we don't even have the same
final window model in mind in the end.

> And with that, I totally disagree with your proposal to essentially regress the design.

But then, what is the point of having the parameter in vout_window_SetFullscreen
if it's used as variable by the underlying window module ? How do you handle the
two following situations ?

+ The user wants to set its preferences to a window-defined screen. Where does
it get defined ? How is it given to the window module ?
+ The user wants to put the current window in fullscreen. Does it uses the
previous setting or should it only be at start when the fullscreen flag is set?

Having the function would have a lot more meaning to me if you had a contextual
menu, but using a menu to move a window doesn't look like the best UX.

What you want to do with this API is not really clear to me in term of UX, I'm
sorry if you find it obvious from your point of view, but I hope it can lead to
better documentation and UX for any end-user of this API.

Regards,
--
Alexandre Janniaux
Videolabs

> Le 14 octobre 2019 23:42:26 GMT+03:00, Alexandre Janniaux <ajanni at videolabs.io> a écrit :
> >Hi,
> >
> >The rationale for having a per-module value is that the
> >would have some meaning only for the modules that report
> >it. Integer-screen name like xinerama ones could be used
> >as string too by having names like xinerama-0, xinerama-1
> >as they are mostly obsolete. Most systems with screens and
> >virtual screens are using strings nowadays because they are
> >easier to use by the end-user.
> >
> >Currently we have vout_window_ReportOutput to report these
> >ids but I'm not sure it's even used by the core and the
> >wayland module is the only producer. Maybe we should add
> >more support for these reports but I see two issues related
> >to their usage:
> >
> >When using audio devices, I tend to think that the user
> >don't really care if they have to change audio output
> >during the playback because the "main" audio device for the
> >common normal user (meaning wasapi/pulseaudio) is already
> >handled by the API of the audio multiplexer below, and
> >changing audio device is an occasional task (or the user
> >didn't really configured it, but that's not our UX problem
> >anymore).
> >
> >In the situation of fullscreen devices, the dual concept is
> >the primary screen one. But in this case primary screen is
> >usually used for different things. For example, the UI are
> >usually started on the primary screen so basically every
> >time the setting "fullscreen-monitor" will have some meaning
> >are the times where you want to set it on a different
> >monitor, or almost, for instance because you have a TV
> >screen connected to the same desktop station. And I don't
> >know if a contextual menu to choose the fullscreen device
> >during the playback is a good idea or not but I think
> >setting it before the playback so that each file opens
> >directly on the correct screen is the use case people want.
> >
> >In this case, you can't use the vout_window_ReportOutput
> >because the window module doesn't exist yet, or for advanced
> >usage you might want to configure it for another module
> >which is not opened and thus reports no fullscreen devices.
> >This is exactly what led us to using Qt as fullscreen
> >provider with QScreen indices at the first place.
> >
> >I advised the added variable because it allows fixing most
> >of the usability issues without having to solve the issues
> >above. Either the UI is the correct provider for the screen
> >name, so it solves for the 99%, or the user is and it solves
> >for the 0.99% left.
> >
> >I really don't think we can get the ideal behaviour before a
> >lot of work both on the core and all the window module, not
> >to mention that there is few documentation on this, although
> >what is there almost correctly specify all the API.
> >
> >Regards,
> >--
> >Alexandre Janniaux
> >Videolabs
> >
> >On Mon, Oct 14, 2019 at 10:05:24AM +0200, Thomas Guillem wrote:
> >>
> >>
> >> On Sun, Oct 13, 2019, at 16:22, Gabriel Luci wrote:
> >> > I assume you mean because the output plugin doesn't necessarily
> >have a window? So isn't this the same as all the "Window properties"
> >settings, which are defined in the same place? Or would you prefer to
> >see this under that "Window properties" section too?
> >>
> >> This option is a string for windows, but it could be an integer on
> >other OSes
> >> Rémi would prefer to have this option in win32/window.c
> >>
> >> vlc_module_begin()
> >>  set_shortname(N_("Win32 window"))
> >>  set_description(N_("Win32 window"))
> >>  set_category(CAT_VIDEO)
> >>  set_subcategory(SUBCAT_VIDEO_VOUT)
> >>  set_capability("vout window", 10)
> >>  set_callback(Open)
> >> vlc_module_end()
> >>
> >> But I'm not sure about it, on Linux, this id would be something like
> >"0:0", so a string too.
> >> Personally, I prefer to have a global settings for all OSses (and
> >maybe do some integer conversion on some implementations).
> >>
> >> >
> >> > On Sun, Oct 13, 2019 at 3:18 AM Rémi Denis-Courmont
> ><remi at remlab.net> wrote:
> >> >> Le sunnuntaina 13. lokakuuta 2019, 6.58.28 EEST Gabriel Luci a
> >écrit :
> >> >> > This will contain the name of the display device to always show
> >> >> > fullscreen video on.
> >> >> > ---
> >> >> > src/libvlc-module.c | 6 ++++++
> >> >> > 1 file changed, 6 insertions(+)
> >> >> >
> >> >> > diff --git a/src/libvlc-module.c b/src/libvlc-module.c
> >> >> > index d3da4e5fb8..163e4deda0 100644
> >> >> > --- a/src/libvlc-module.c
> >> >> > +++ b/src/libvlc-module.c
> >> >> > @@ -314,6 +314,11 @@ static const char *const
> >ppsz_align_descriptions[] =
> >> >> > #define FULLSCREEN_LONGTEXT N_( \
> >> >> > "Start video in fullscreen mode" )
> >> >> >
> >> >> > +#define FULLSCREEN_MONITOR_TEXT N_("Display device for
> >fullscreen video")
> >> >> > +#define FULLSCREEN_MONITOR_LONGTEXT N_( \
> >> >> > + "In a multiple monitor configuration, you can specify the
> >device name "
> >> >> > \ + "of the display that you want the fullscreen video window to
> >open
> >> >> > on." ) +
> >> >> > #define VIDEO_ON_TOP_TEXT N_("Always on top")
> >> >> > #define VIDEO_ON_TOP_LONGTEXT N_( \
> >> >> > "Always place the video window on top of other windows." )
> >> >> > @@ -1641,6 +1646,7 @@ vlc_module_begin ()
> >> >> > add_bool( "fullscreen", false, FULLSCREEN_TEXT,
> >FULLSCREEN_LONGTEXT,
> >> >> > false ) change_short('f')
> >> >> > change_safe ()
> >> >> > + add_string( "fullscreen-monitor", "", FULLSCREEN_MONITOR_TEXT,
> >> >> > FULLSCREEN_MONITOR_LONGTEXT, false ) add_bool( "embedded-video",
> >1,
> >> >> > EMBEDDED_TEXT, EMBEDDED_LONGTEXT, true )
> >> >> > add_bool( "xlib", true, "", "", true )
> >> >>
> >> >> The values depend on the output plugin, so thus does not belong as
> >a global
> >> >> setting, just like the audio output device.
> >> >>
> >> >> --
> >> >> レミ・デニ-クールモン
> >> >> http://www.remlab.net/
> >> >>
> >> >>
> >> >>
> >> >> _______________________________________________
> >> >> vlc-devel mailing list
> >> >> To unsubscribe or modify your subscription options:
> >> >> https://mailman.videolan.org/listinfo/vlc-devel
> >> > _______________________________________________
> >> > vlc-devel mailing list
> >> > To unsubscribe or modify your subscription options:
> >> > https://mailman.videolan.org/listinfo/vlc-devel
> >
> >> _______________________________________________
> >> vlc-devel mailing list
> >> To unsubscribe or modify your subscription options:
> >> https://mailman.videolan.org/listinfo/vlc-devel
> >_______________________________________________
> >vlc-devel mailing list
> >To unsubscribe or modify your subscription options:
> >https://mailman.videolan.org/listinfo/vlc-devel
>
> --
> Envoyé de mon appareil Android avec Courriel K-9 Mail. Veuillez excuser ma brièveté.

> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel