<div dir="ltr">If you want to avoid the extra round trip, go down a level in XCB and send the get_geometry and frame_extents (and maybe whatever else) requests before looking for a response. This is the point of the cookie mechanism: don't wait around for data before sending unrelated requests. </div> <div class="gmail_quote"><div dir="ltr">On Sun, Jul 5, 2015 at 5:08 AM Uli Schlachter <<a href="mailto:psychon@znc.in">psychon@znc.in</a>> wrote: </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi, Am 04.07.2015 um 18:12 schrieb Cosmin Apreutesei: >>> - xcb_get_geometry always returns the initial dimensions even after >>> the window gets resized (what gives?) >> >> That makes no sense at all. Do you have a small example on this? >> > > Hi again Uli, thanks for answering. > > It's a Lua backend for a cross-platform windowing lib[1] so I can't easily > make C examples. > > Anyway, your skepticism made me look deeper and I finally found the > problem. I was enabling too many bits on the event mask, including > XCB_EVENT_MASK_RESIZE_REDIRECT which was causing the described behavior. I > was confident I did nothing wrong because of a note in the pyglet source > code saying the same thing about XGetGeometry, but apparently the author > fell into the same trap by coincidence. Sorry for the noise and thanks for > the help. Hey, that's a nice trick to make your window un-resizable. I have to remember that. :-) > About the 4.1.5 section of the ICCCM, I can do it that way but first I want > to ask: wouldn't the following stateless piece work too? > > frame_rect(win) { > x, y = xcb_translate_coords(win, screen.root, 0, 0) > geom = xcb_get_geometry(win) > w, h = geom.width, geom.height > w1, h1, w2, h2 = xcb_frame_extents(win) > return x - w1, y - h1, w + w2, h + h2 > } Yup, I guess this would work. However, ICCCM tries to avoid a round-trip. I don't think worrying about a single round-trip being slow is that important these days. So yeah, the above code is fine with me. > This assumes the wm gives frame extents -- are there known wms that don't? [...] Any WM that doesn't implement EWMH or just doesn't implement this part of EWMH. Any WM that doesn't have window decorations. (dwm? i3wm? awesome?) Cheers, Uli P.S.: I guess I should eventually fix awesome to implement this bit of EWMH... -- 99 little bugs in the code 99 little bugs in the code Take one down, patch it around 117 little bugs in the code -- @irqed _______________________________________________ Xcb mailing list <a href="mailto:Xcb@lists.freedesktop.org" target="_blank">Xcb@lists.freedesktop.org</a> <a href="http://lists.freedesktop.org/mailman/listinfo/xcb" rel="noreferrer" target="_blank">http://lists.freedesktop.org/mailman/listinfo/xcb</a></blockquote></div>