On 28/12/2025 20:10, ichthyo wrote:
My hypothesis is thus far, that when running under X11,
the coordinate system is somehow different...
On 28/12/2025 20:57, ichthyo wrote:
And I can confirm my hypothesis now.
...but unfortunately I did not get any further. :-(
I looked into the transformation matrix used by Cairo,
which can be retrieved with cairo_get_matrix()
This indicates that Cairo draws unscaled in all cases.
For Wayland the origin offset is (0.5, 0.5)
while for X11 it is reported as (0,0)
I tried to set various offsets, like ±0.5 or
±window-coordinates (those strange large values
reported by the FLTK window.
But with no avail.
The source code for Fl::cairo_make_current() is implemented
completely different for the Wayland case and the X11 case.
And quite some strange stuff for the X11 case, while the
Wayland case looks straight forward.
Anyway, I'm running out of ideas now and need to take a break.
If anyone has further ideas, please go ahead...
Maybe we should try to ask someone who has a better understanding?
Any ideas whom to ask?
Maybe posting on Stackoverflow, what do you mean?
Btw: the same technique works perfectly well for the WidgetDial.
So presumably our problems are related to that grapic being
in a pop-up window.
@Will on a related note: I am quite sure that these problems
are linked to the technique of accessing the Cairo context.
Thus there seems to be no connection to my patch, which only
attempts to use the right code path in some corner cases,
like when FLTK was built without Cairo bridge.
Thus IMHO, since Jesper's patch is already on master,
my patch could go in as well, since its more like a
little amendment to Jesper's work
-- Hermann
_______________________________________________
Yoshimi-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/yoshimi-devel
