On Fri, Apr 12, 2019 at 12:23:33PM -0400, Okan Demirmen wrote: > On Fri 2019.04.12 at 11:32 -0400, Charles wrote: > > On Fri, Apr 12, 2019 at 10:50:54AM -0400, Okan Demirmen wrote: > > > On Fri 2019.04.12 at 10:15 -0400, Charles wrote: > > > > On Fri, Apr 12, 2019 at 09:26:36AM +0200, Benjamin Baier wrote: > > > > > On Thu, 11 Apr 2019 22:04:07 -0400 > > > > > Charles <char...@cdaniels.net> wrote: > > > > > > > > > > > Hello list, > > > > > Hello > > > > > > > > > > > There seems to be an issue with the Scribus port when combined with > > > > > > CWM. > > > > > > Namely, the "properties" window/dialog can only be opened once. > > > > > > After > > > > > > being closed, it cannot be opened again. Restarting the application > > > > > > does > > > > > > not fix the problem. > > > > > > > > > > > > This issue is specific to CWM, I am not able to reproduce under > > > > > > FVWM. > > > > > > > > > > > > I have reproduced this issue under OpenBSD current on AMD64, and on > > > > > > OpenBSD 6.4 on macppc (iBook G4). In both cases, using Scribus 1.4.6 > > > > > > installed via pkg_add. > > > > > > > > > > > > I found a previous issue that affected the way CWM and Qt interact, > > > > > > which was > > > > > > posted[1] under the subject line "calmwm mouse stuck inside of > > > > > > window" over on > > > > > > misc@. Okan and I synced up off list and determined it was due to > > > > > > CWM's > > > > > > implementation of EWMH interacting improperly with Qt. > > > > > Seen this one before, too. > > > > > > Yes, in fact this has to do with cwm's insistence of warp'ing the mouse; > > > a lot of cwm's behaviours get in the way of ewmh - or maybe one could > > > say the other way around. Changing cwm various behaviours are > > > challenging because wm's are super personal and no one ends up happy > > > (including me); as this is my experience at least. /end mini-rant. One > > > might also say Qt behaves differently as well...anyway. > > > > <rant> > > > > It is unfortunate that this type of thinking (referring to the general > > philosophy of OpenBSD / CWM / suckless.org / et. al.) seem to swim so > > hard against the current. I think it is not so unreasonable to expect my > > WM to act on my behalf and not yank windows around without my say so, > > and so on... unfortunately it would seem that I/we are in a minority on > > that point. > > > > </rant> > > > > Do you think there is a strong possibility this is EWMH related? If the > > EWMH event handler is disabled can the issue be reproduced still? I plan > > to spend some time testing this more rigorously over the weeknd; I can > > add CWM w/ EWMH patched out to my matrix of test configurations, > > assuming I can figure out how to switch it off easily. As I recall it > > had a pretty well defined entry point, but I haven't read the code in a > > while. > > Disable all the pointer warp'ing, or EWMH (active window hint) - either > will "fix" your issue in the short term.
@Okan It turns out that this actually did not fix the issue. I may have done it wrong though... In 'xev_handle_clientmessage(XEvent *ee)' in xevents.c I commented out everything that called any xu_ewmh. I also commented out 'xu_ewmh_net_desktop_names(sc);' from xev_handle_propertynotify(XEvent *ee). Finally, I commented out the body of 'client_ptrwarp(struct client_ctx *cc)' in client.c. I will still able to reproduce the issue as discussed previously. @Stuart > If you're trying to ascertain whether some problem is specific to > cwm, you should compare with another non-reparenting window manager > (there's a list of common ones at > https://en.m.wikipedia.org/wiki/Re-parenting_window_manager - for > example dwm is probably the most common one) as well as a more > standard reparenting wm. This should give you a better idea where the > problem lies and give you more details that you can report upstream > if necessary. I tested under dwm on OpenBSD 6.4 on macppc and was not able to re-produce the issue. Unfortunately it would appear that this is cwm specific. I have yet to test this under cwm on Linux (or any WMs aside from fvwm, cwm, and dwm). I will report back once I have tested more configurations, but I think this runs deeper than just EWMH weirdness. Maybe someone with some qt experience can chime in? I would also be happy to be point-of-contact for upstream if we want to see if they have any insight. ~ Charles
