some VALIDATE_WIN_COPY calls should be VALIDATE_WINVALID_COPY, but haven't exhaustively looked for these. --10/02/99 gjb ------- Out of color errors need to be reported better. ------- closing the Fvwm pager causes a deadlock that makes Scwm hang; kill -1 twice wakes it up, but need real fix. --tlau ------------- Race conditions cause scwm to hang when talking to fvwm2 modules when, e.g., recapturing. May have just been forgetting the O_NONBLOCK flag... --09/17/98 gjb ------- dialog popups that grab the keyboard focus should return focus better after they are unmap (e.g., with tlau's fugu's search dialog, and escape aborting out of it). -> how get key focus -> why keybindings not work after the Escape is pressed -> focus-window-with-pointer should work... ------- using (gtk gtk) before (app scwm gtk) causes a premature exit. How can we force app scwm gtk to be called when (gtk gtk) module is loaded? --10/02/99 gjb ------- XEmacs pixmap icon doesn't survive after a restart (James Achtung Crippe) --09/15/99 gjb ------- > btw, you might check for dl support before building utilities/pidproperty/, > at least obsd didn't recognize the RTLD_GLOBAL flag on dlopen, had to > replace it with DL_LAZY, at least it compiled, not sure if it works. had > to remove -ldl from the Makefile too. (aren't there some dlopen tests > done in configure already ?) Bob Pepin bob@sendar.prophecy.lu -------- kept-on-top windows cannot be restacked relative to each other. I tried finding some quick fixes but they cause segfaults, thus this is something to fix by reconsidering all this code and implementing true layers functionality. ------------- Graeme.Cross@sci.monash.edu.au reports problems using ApplixWare's windows and Scwm (perhaps related to bitmaps that Applixware puts on the windows) ------------- Squashed titlebar feature: it'd be nice if they could be squashed to the right, or centered, instead of always only left-justified. ------------- emacs lisp files get installed under / if no emacs (from forcer --09/23/98 gjb) ------------- 1. I pop up an xterm 2. I do a horizontal-toggle-maximize 3. I change to a smaller font. 4. The window automatically resizes due to 3, preserving the number of rows & columns. 5. I do a horizontal-toggle-maximize again to bring it back to normal size. 6. The window resizes to many more rows and columns than it originally had. It ends up about the same size in pixels, though. Evidentally toggle-maximize isn't differentiating between windows sized in characters & windows sized in pixels. Shouldn't toggling back to the original size go back to the orignal number of rows & columns, not the original size in pixels? If I recall correctly, that's what fvwm does. hstein - 7/16/98 ------------- Build bugs --04/23/99 gjb -lXpm was not on my link line when I linked on Hydra.cs.mu.oz.au -- why? gcc didn't find -lstdc++ on hydra, either, and I needed to add -L/usr/lib/gcc-lib/i486-linux/egcs-2.91.60 to get it. (Are these debian issues?) ------------- Test with bad image path-- lots of things could be done better when images are missing. ------------- Visual bugs in boundaries of windows with some configurations of options. ------------- icon-box not honoured correctly? ------------- popup-menus attached to decorations come up centered over the click, instead of flush with the window edge -- use of popup-menu-from-decoration is a bit hackish --------------- xdvorak's change mapping should cause all keybindings to be reset and new keycodes to be gotten (waiting for event rewrite) --------------- Windows still aren't positioning correctly for me at startup. They're starting too low (when positioned with -0, as in: xterm -j -sb -sl 128 -si -geometry 82x48-0-0 -T $XTERMNAME & xterm -j -sb -sl 128 -si -geometry 80x24+0-0 -T $XTERMNAME & The 1st xterm also ends up a little to far to the right (looks like it's off by the window border). This is with "guile 1.3a" (snap 980630), & a cvs update this morning of scwm. --hjstein ------------- I've noticed this off-and-on with 0.6, and just noticed again with 0.7: My xterms have #:focus 'sloppy set; sometimes (probably about one time in 25), when I have a pair of partially overlapping windows and I move the cursor quickly from the root window, crossing one of the xterms (which held the sloppy focus), and landing in the other xterm, the new window does not gain focus. I have to take the cursor off of the window in which I desire focus and bring it back in, crossing the border slowly. Ken Pizzini ------------- - Another window shade problem: clients that resize themsleves while shaded make the wm lose. [ reported by Oleg Tihnov ] ------------ - Click to focus bugs 1a. At startup usually a window, which I don't want given focus, gets focus. 1b. The same is true on a restart. Except sometimes no window gets focus at all, which makes life difficult for `next'. 2. When I quit a X app, the focus is usually given to a window I don't want to have focus. Jens ------- callbacks should run from a new dynamic root. [ this would prevent them from calling continuations that escape outside the callback handler - the code would lose big on this -ms] GONE??? (but didn't note resolving them here) ------- --01/09/00 gjb By-window-resource group menus should auto-select the first window in the sub-menus (they currently require a motion event for the menu item to get highlighted). ------------- --01/09/00 gjb Incorrect usage of keyword arguments by a scheme function fails silently. Both wrong keyword names and missing arguments should be reported. ------------- ---------------------------------------------------------------------- RESOLVED: ---------------------------------------------------------------------- Sticky windows do not de-iconify properly. --01/09/00 gjb ------- --01/09/00 gjb There seems to be a bug in window size (or allowable size) communication between scwm & applications. For example, bring up a netscape window. Maximize its height. It very nicely stretches from the top of the window to the bottom. Now start giving it lots of Alt-Ns to pop up new netscape windows. Netscape always starts up new windows the same size as the netscape window last created or adjusted in size. So, all the windows being poped up should be the same size. However, each window poped up ends up one pixel larger in height. --Harvey Stein ------- Changing desktops on elements GWS machines (as tzoompy) crashes Scwm; I can't reproduce as gjb NLR (No longer reproducible)--09/29/99 gjb ------- start-x2x-ita causes scwm to grow wildly out of control until that new x2x process is killed. ConfigNotify fighting + guile leak? --09/29/99 gjb ------- *preferences-use-notebook* set to #f causes a seg fault in the gc from guile-gtk code, it seems. --09/13/99 gjb Res'd -- Vollmer's patch to guile-gtk --09/29/99 gjb ------- xload windows (among others, probably) move and resize on restarts -- gravity related? Res'd --07/17/99 gjb ------------- wildcard-matcher in wininfo doesn't permit distinguishing based on title, class, and resource name separately. This is essential for window styles to do the right thing (a misfeature compared to fvwm2) ---------------------------------------------------------------------- Really smart placement (i.e., clever placement) has an off my one bug. Try starting a couple of xterms in the viewport one below the home viewport, then move back to the home viewport -- you can see one row of pixels at the bottom of the screen. --04/14/99 gjb Res'd--07/17/99 gjb ---------------------------------------------------------------------- make-color can return #f, we should have make-closest-color that is guaranteed to work and return the closest color to the requested one. ------------- Hooks which take wrong number of params cause seg faults! --07/04/98 gjb ---------------------------------------------------------------------- windows having a different border-width than the default (6) are placed incorrectly if they have SE gravity. They are several pixels [(6-border_width)*2] off the screen-border. This is due to the fact that `PlaceWindow' is called before the evaluation of window-styles. `PlaceWindow' needs to be called if the border-width changed (from `set_border_width_x', or `SetupFrame', or whatever). It would be nice if the call to `PlaceWindow' before the window is properly styled could then be removed, but I don't know what this could break. Robert Bihlmeyer Robert verified this has been fixed -- 09/26/98 jtl ------------ Changing title-height in the decor (e.g., by changing the font) doesn't affect new windows using that style ------------- xterm-s started from scwm die when scwm dies.... why? see scwm-system in base.scm -- the terminal process group was getting the SIGINT from the controlling tty --09/17/98 gjb ------------- gjb gets this on recapture [Scwm][ScwmErrorHandler]: <> Request 54, Error 4, EventType: 2 XFreePixmap This could be due to the recapture code losing track of the pixmap due to some change I made in the pixmap code --11/08/97 gjb Not reproducible any longer --08/02/98 gjb - Animated window-shade bug, reported by Sam Steingold , exploit confirmed by MS, but I have no clue how to fix it right now: Open 2 windows, one (1) partially obscuring the other's (2) title bar. Double click on the (1)'s title bar, shading it. You can see that the portion of the (2)'s title bar with was obscured before but it not obscured now is not repainted. Also, if you move the mouse towards (2) while (1) is being (animatedly ==> with delay) shaded, so that the (1) is "pulled from under the mouse pointer", leaving it on (2), (2) will *not* have the focus, although it should as per sloppy focus. --03/10/98 gjb; XSync was discarding events, and we needed a CoerceEnterNotifyOnCurrentWindow() to restore proper focus - Jim Blandy's patch to load boot-9 for newer guile versions is problematic; it can't tell wether or not we are running under an older guile which has already loaded boot-9, resulting in a segfault. - MS [fixed for now - MS] - buggy icon handling everywhere -- needs thorough testing -- 2-20-97 mstachow should initially focus/hilight some window (also a bug in fvwm2) -- Fixed by adding CoerceEnterNotifyOnCurrentWindow() Memory allocation errors can occur if guile was built with Rx support; autoconf sets this option if it finds a librx.{a,so} in the paths it checks. --11/08/97 gjb Passing a bad window from guile can cause scwm to dump core in ensure_valid(); MS says fixed ~ 11/1/97 functions should be renamed to end with _x -- This looks done... is it not? --11/08/97 gjb (get-window) acts wierdly when a window needs to be selected, or when executed via scwmsend's property mechanism -- was a problem with DeferExecution(); not sure if it's completely right now, but seems to work in the cases I've tried -- 10/25/97, gjb there needs to be a way to remove mouse bindings; unbind_key added 10/24/97, gjb start interactive moves and resizes right, so that initial mouse motion is not lost. ~ 10/15/97, MS Oleg Tihonov 's bug re: bind-mouse and unbind-mouse. bind-mouse grabbed for all windows unnecessarily, and unbind-mouse did not permit `1' to mean moust button 1 --gjb 11/27/97