[cairo] xcb surface example?

[Bryce Harrington]

On Sat, Jul 04, 2015 at 03:51:25PM +0200, Uli Schlachter wrote:
> Am 29.06.2015 um 21:53 schrieb Bryce Harrington:
> > On Fri, Jun 26, 2015 at 08:09:36PM +0200, Uli Schlachter wrote:
> >> Am 26.06.2015 um 13:31 schrieb Andreas Lobinger:
> >>> Hello colleagues,
> >>
> >> Hi,
> >>
> >>> is there a good, known example of the cairo XcbSurface available? On the
> >>> http://cairographics.org/examples/ page there is a link to
> >>> git clone git://git.thisnukes4u.net/cairo-xcb-demo.git
> >>> which seems to be outdated.
> >>
> >> What exactly do you need? There is mostly just a single useful function in
> >> cairo-xcb. :-)
> >>
> >> Anyway, attached is a quick example that I hacked together. Hopefully it can
> >> help you.
> > 
> > I've added this as a new link at the end of our cookbook:
> > 
> >     http://cairographics.org/cookbook/
> > 
> > ...and removed the linkrot from our examples page.
> > 
> > Uli, would you mind specifying what license should be listed for your
> > code?  (Fwiw, the blur.c code appears to be using an MIT-ish license.)
> 
> Feel free to consider this WTFPL 2.0 (or any other license you like). I
> seriously doubt this would even be covered by copyright at all. I still don't
> think this is really a good example.
> 
> Hm... The find_visual() function originally comes from awesome
> (awesome.naquadah.org), so I guess this makes this code GPLv2. Still, I don't
> think this function is covered by copyright either.

Looking at awesome/draw.c's draw_find_visual() function, the code is
similar in spirit but quite different in implementation.  And given that
the find_visual() is so short compared with the rest of the program, I
doubt its licensing matters too much here.

So, I'll use the MIT license from blur.c for this for consistency.  Feel
free to change.

> [...]
> >> #include <xcb/xcb.h>
> >> #include <cairo-xcb.h>
> >> #include <stdio.h>
> >> #include <stdlib.h>
> >>
> >> static xcb_visualtype_t *find_visual(xcb_connection_t *c, xcb_visualid_t visual)
> >> {
> >> 	xcb_screen_iterator_t screen_iter = xcb_setup_roots_iterator(xcb_get_setup(c));
> >> 	
> >> 	for (; screen_iter.rem; xcb_screen_next(&screen_iter)) {
> >> 		xcb_depth_iterator_t depth_iter = xcb_screen_allowed_depths_iterator(screen_iter.data);
> >> 		for (; depth_iter.rem; xcb_depth_next(&depth_iter)) {
> >> 			xcb_visualtype_iterator_t visual_iter = xcb_depth_visuals_iterator(depth_iter.data);
> >> 			for (; visual_iter.rem; xcb_visualtype_next(&visual_iter))
> >> 				if (visual == visual_iter.data->visual_id)
> >> 					return visual_iter.data;
> >> 		}
> >> 	}
> >>
> >> 	return NULL;
> >> }
> >>
> >> int main()
> >> {
> >> 	xcb_connection_t *c;
> >> 	xcb_screen_t *screen;
> >> 	xcb_window_t window;
> >> 	uint32_t mask[2];
> >> 	xcb_visualtype_t *visual;
> >> 	xcb_generic_event_t *event;
> >> 	cairo_surface_t *surface;
> >> 	cairo_t *cr;
> >>
> >> 	c = xcb_connect(NULL, NULL);
> >> 	if (xcb_connection_has_error(c)) {
> >> 		fprintf(stderr, "Could not connect to X11 server");
> >> 		return 1;
> >> 	}
> >>
> >> 	mask[0] = 1;
> >> 	mask[1] = XCB_EVENT_MASK_EXPOSURE;
> >> 	screen = xcb_setup_roots_iterator(xcb_get_setup(c)).data;
> >> 	window = xcb_generate_id(c);
> >> 	xcb_create_window(c, XCB_COPY_FROM_PARENT, window, screen->root,
> >> 			20, 20, 150, 150, 0,
> >> 			XCB_WINDOW_CLASS_INPUT_OUTPUT,
> >> 			screen->root_visual,
> >> 			XCB_CW_OVERRIDE_REDIRECT | XCB_CW_EVENT_MASK,
> >> 			mask);
> >> 	xcb_map_window(c, window);
> >>
> >> 	visual = find_visual(c, screen->root_visual);
> >> 	if (visual == NULL) {
> >> 		fprintf(stderr, "Some weird internal error...?!");
> >> 		xcb_disconnect(c);
> >> 		return 1;
> >> 	}
> >> 	surface = cairo_xcb_surface_create(c, window, visual, 150, 150);
> >> 	cr = cairo_create(surface);
> >>
> >> 	xcb_flush(c);
> >> 	while ((event = xcb_wait_for_event(c))) {
> >> 		switch (event->response_type & ~0x80) {
> >> 		case XCB_EXPOSE:
> >> 			/* Should check if this is the last expose event in the
> >> 			 * sequence, but I'm too lazy right now...
> >> 			 */
> 
> The above could be implemented by adding
> 
>   if (((xcb_expose_event_t *) event)->count != 0)
>      break;
> 
> (If the field is non-null, at least that many more expose events follow
> containing other rectangles that need to be redrawn. By checking for zero, this
> does less redraws.)
> 
> >> 			cairo_set_source_rgb(cr, 0, 1, 0);
> >> 			cairo_paint(cr);
> >>
> >> 			cairo_set_source_rgb(cr, 1, 0, 0);
> >> 			cairo_move_to(cr, 0, 0);
> >> 			cairo_line_to(cr, 150, 0);
> >> 			cairo_line_to(cr, 150, 150);
> >> 			cairo_close_path(cr);
> >> 			cairo_fill(cr);
> >>
> >> 			cairo_set_source_rgb(cr, 0, 0, 1);
> >> 			cairo_set_line_width(cr, 20);
> >> 			cairo_move_to(cr, 0, 150);
> >> 			cairo_line_to(cr, 150, 0);
> >> 			cairo_stroke(cr);
> >>
> >> 			cairo_surface_flush(surface);
> >> 			break;
> >> 		}
> >> 		free(event);
> >> 		xcb_flush(c);
> >> 	}
> >> 	cairo_surface_finish(surface);
> 
> This is missing a call to
> 
>     cairo_surface_destroy(surface);

Got it, thanks.
Bryce
 
> >> 	xcb_disconnect(c);
> >> 	return 0;
> >> }
> 
> Cheers,
> Uli
> -- 
> Who needs a ~/.signature anyway?