urban@spp2.UUCP (Michael Urban) (07/02/88)
If I use the Xt toolkit to create a Transient pop-up shell, AWM insists on performing "rubber-banding" (calling AskUser), because the USPosition and USsize hints are not both set; however, the Shell widget does not set these hints unless a "geometry" resource is supplied for the pop-up shell, even if size and location information is supplied (or, as in my case, location is supplied and the size is determined by the Shell's contents). Which of the three participants (awm, Shell(Xt), my application) should be changed? -- Mike Urban ...!trwrb!trwspp!spp2!urban "You're in a maze of twisty UUCP connections, all alike"
dheller@cory.Berkeley.EDU (Dan Heller) (07/05/88)
In article <350@algol.spp2.UUCP> urban@spp2.UUCP (Mike Urban) writes: >If I use the Xt toolkit to create a Transient pop-up shell, >AWM insists on performing "rubber-banding" (calling AskUser), >because the USPosition and USsize hints are not both set; >Which of the three participants (awm, Shell(Xt), my application) >should be changed? > Mike Urban > ...!trwrb!trwspp!spp2!urban I fixed this by setting the override_redirect flag to true in the window attributes structure. Awm works fine in this context. However, there is still a problem with destroying this shell. In the transient shell that I created, I placed a number of widgets, one of which is a "quit" command button which merely calls XtDestroyWidget() on the transient shell widget. The next call to XtNextEvent tries to do the second phase of the destruction process and core dumps deep within Xlib ... The only work around I can come up with is not destroying the widget, but unmapping it instead. This is bad because I can't free all the memory and resources it consumes. the question is, what's wrong with destroying a transient shell? Dan Heller <island!argv@sun.com>
dshr@SUN.COM (David Rosenthal) (07/05/88)
It seems pretty clear to me that AWM is wrong. Asking the user to rubber-band out the space for a transient window, which is intended to be something like a dialog box, has to be wrong. Would you want to be interrupted in order to stroke out the space in which an application was to ask you "OK to exit?" I cannot understand why any window manager would ask the user to interactively size any WM_TRANSIENT_FOR window, irrespective of the setting of the USPosition or USSize hints. I would be prepared to forbid this in the Inter-Client Communication Conventions Manual. Using override-redirect on Transient pop-ups is definitely wrong. David.
ellis@AUDI.SIEMENS.COM (Ellis Cohen) (07/05/88)
> In article <350@algol.spp2.UUCP> urban@spp2.UUCP (Mike Urban) writes: > >If I use the Xt toolkit to create a Transient pop-up shell, > >AWM insists on performing "rubber-banding" (calling AskUser), > >because the USPosition and USsize hints are not both set; > > >Which of the three participants (awm, Shell(Xt), my application) > >should be changed? > > Mike Urban > > ...!trwrb!trwspp!spp2!urban > > I fixed this by setting the override_redirect flag to true in the > window attributes structure. Awm works fine in this context. Please do not set encourage application writers to set override_redirect on transient windows. Override_redirect windows are meant to be invisible to window managers. They should only be used for extremely short-lived windows like pop-up menu. Transient windows typically have longer lifetimes. Many window managers will provide decorations for transient windows so that the user can move them around. Setting override_redirect defeats that goal. My gut feeling is that awm should put the transient window up at the size and position requested. Ellis Cohen Siemens RTL
jkh@ardent.UUCP (07/06/88)
It's not apparent to me that awm *is* asking to confirm transient popup windows. xmh, for example, pops up confirm boxes all over the place (for me) under awm without any rubberbanding. That's not to say that there isn't a subtle bug lurking in awm somewhere (for which I will look) that causes this not to happen on occasion, but just to set the record straight, it's not (and never has been) awm's intention to interfere with transient windows. I don't need an ICCCM spec to know that. Jordan
obrien@anpiel.aero.org (Michael O'Brien) (07/06/88)
In article <8807051604.AA02096@devnull.sun.com> dshr@SUN.COM (David Rosenthal) writes: > >Using override-redirect on Transient pop-ups is definitely >wrong. I'm prepared to believe this, but I think it interesting that in X.V11R2, the "bitmap" program does exactly this. I recently reamed that program to provide some additional capability (don't ask for it - it's specific work for a project and useless in general) and had a whole bunch of trouble until I turned that bit off. Of course, it also doesn't say TRANSIENT anything. I have a suspicion that the "bitmap" code is very early X stuff. -- Mike O'Brien obrien@aerospace.aero.org {sdcrdcf,trwrb}!aero!obrien