Re: Tooltips progress

[Tim Janik <timj imendio com>]

On Thu, 1 Jun 2006, Michael Natterer wrote:

> On Wed, 2006-05-31 at 22:08 +0200, Kristian Rietveld wrote:
>
> > Currently, we are using the following query-tooltips signal:
> >
> >   gboolean (*query_tooltip)     (GtkWidget           *widget,
> >                                  gint                 x,
> >                                  gint                 y,
> >                                  GtkTooltip          *tooltip);
> >
> > But if a user sets a custom using gtk_widget_set_custom_window(), we don't
> > have to pass a GtkTooltip around.  So currently I just set the tooltip
> > field to NULL, so the user knows he has to use his own custom tooltip
> > window (and we assume he has a reference to that).  Is this okay or do
> > we want to change this?  We could also move the _set_custom_window()
> > functions into GtkTooltip for example.
>
> I would very much appreciate if the GtkTooltip object also kept track
> of the custom window. IMHO it makes no sense to handle this differently.

the widget is meant to do that:
void            gtk_widget_set_tooltip_window   (GtkWidget      *widget,
                                                 GtkWindow      *custom_window);
GtkWindow*      gtk_widget_get_tooltip_window   (GtkWidget      *widget);
for reasons outlined in a previous reply from me to kris.

> gtk_tooltip_set_markup() vs. gtk_tooltip_set_window()
>
> simply feels much better than the asymmetry in
>
> gtk_tooltip_set_markup() vs. gtk_widget_set_tooltip_window()

well, you'd not use GtkTooltip at all if you used 
gtk_widget_set_tooltip_window() before. in that way,
the "asymmetry" simply reflects your usage.

> The current approach even limits the new tooltip system's use cases,
> since you have to set the custom window *outside* the callback. There
> is no way to decide *in* the callback if a standard tooltip is
> sufficient, or if a custom window is needed.

i don't think it'd be clear that/if the custom window will be available
in the GtkTooltip object again upon the next ::query-tooltip()
emission. especially since users are not supposed to access GtkTooltip
objects/structures in anyway *outside* of ::query-tooltip(), as outlined
originally:
  http://mail.gnome.org/archives/gtk-devel-list/2006-April/msg00010.html

also, passing GtkTooltip as NULL is meant to be as a clear indication
that the user has to tinker with gtk_widget_get_tooltip_window()
instead of the tooltip object. this avoids ambiguities like:

static gboolean
query_tooltip (GtkWidget           *widget,
               gint                 x,
               gint                 y,
               GtkTooltip          *tooltip)
{
  gtk_tooltip_set_markup (tooltip, "See Me?");
  g_object_new (GTK_TYPE_BUTTON,
                "parent", gtk_widget_get_tooltip_window(),
                "label", "Can You Click Me?",
                NULL);
  gtk_tooltip_set_icon (tooltip, messup_pixbuf);
}

what's the tooltip code supposed to do here?

and yes, the API basically enforces that you decide per widget if
an own tooltip window is needed or not. but i don't think that is
a strong limitation, we have been discussing usage cases quite some
in thise thread and nothing required to defer this decision into
the callback (and no existing tooltip code i know of requires this
either).
so giving the benefits in API (and implementation) this provides,
i think it is a reasonable limitation to make.

> Another problem is the assymetry in getting the relevant data to the
> callback. With markup, the GtkTooltip can be queried for the markup,
> if it's a custom window, you have to get it to the callback somehow.

Kris just forgot the getter for the custom window on the widget,
so signal handler user_data is left alone.

> In your example code it's passed as user_data, but user_data is
> often not available since it's needed for other things, so you
> have to add a "GtkWidget *tooltip_window" to the "other thing".
> If "other thing" is e.g. the application toplevel window, which
> of the sub-widget's tooltips should it keep? All of them? People
> will in many cases end up attaching the tooltip window to the
> widget.
>
> I also fail to see why the "has-tooltip" property has to be
> set to TRUE, even tho a tooltip window was set on the widget.

i think this can be come by if we implement:
- gtk_widget_set_tooltip_window() should automatically set:
  GtkWidget::has-tooltip = (custom_window != NULL ||
                            GtkWidget::tooltip-markup != "");
- setting GtkWidget::tooltip-markup should automatically set:
  GtkWidget::has-tooltip = (custom_window != NULL ||
                            GtkWidget::tooltip-markup != "");
right?

> ciao,
> --mitch

---
ciaoTJ