toml@Solbourne.COM (Tom LaStrange) (10/14/89)
The subject says it all. Is an "xplaces" program possible?
I am attempting to put xplaces functionality into a window manager and
am running into some problems. The window manager knows a lot of
information about client windows: state, icon position, geometry, etc.
When the window manager is asked to dump out the information to
something that could be used as an .xinitrc file, it would be nice if
all this state information could be saved also.
I can't figure out a standard way to do this.
The window manager can start with the WM_COMMAND property but cannot
really modify it in any way. Not all clients understand the same
options so the wm can't even add something as simple as a -geometry
option to the WM_COMMAND string without breaking clients that don't
understand -geometry. XView clients are a perfect example, all the
options they recognize are Sunview like rather than Xt like.
What I almost need is a standard -ignore "some string here" option that
all clients will ignore. This string could then be looked for in the
WM_COMMAND property when the window manager goes to reparent the
window. The window manager could parse the string for any options to
restore the client to the state from which it was saved.
Have I completely missed the boat on something here? Is there an easy
standard way to do this that I've overlooked?
This is going to be the first tool that a lot of Sunview users are
going to ask for and it's not going to go over well if it only works
with half the clients out there.
--
Tom LaStrange
Solbourne Computer Inc. ARPA: toml@Solbourne.COM
1900 Pike Rd. UUCP: ...!{boulder,nbires,sun}!stan!toml
Longmont, CO 80501janssen@holmes (Bill Janssen) (10/16/89)
In article <2779@ninja.Solbourne.COM>, toml@Solbourne (Tom LaStrange) writes: >What I almost need is a standard -ignore "some string here" option that >all clients will ignore. This string could then be looked for in the >WM_COMMAND property when the window manager goes to reparent the >window. The window manager could parse the string for any options to >restore the client to the state from which it was saved. I think it would be even more useful to have an arbitrary "-property P=D" type of switch, defining properties which all conforming clients would agree to place on their top-level windows. Bill -- Bill Janssen janssen.pa@xerox.com (415) 494-4763 Xerox Palo Alto Research Center 3333 Coyote Hill Road, Palo Alto, California 94304
kjx@comp.vuw.ac.nz (R James Noble) (10/16/89)
I have a trivial program that traverses the window tree, and prints out the WM_COMMAND and WM_CLIENT_MACHINE property for each window, in a form suitable for execution by a shell. It also attempts to find suitable values for the -geometry flag, which it prints out. It seems that most applications upon getting multiple -geom flags ignore all but the last, so I can print the command, and also print my calculated geometry and all works as expected... for example: xexec offramp.comp.vuw.ac.nz xload -geometry 100x100-120+10 \ -geometry 112x80+2+19 There are two problems I have found: firstly, I currently treat icons as uniconified windows (finding the unmapped window with WM_COMMAND set), and secondly, with certain applications (such as an xterm with scrollbar set) the window size I calculate (window size divided by resize increment) does not give the correct geometry specification for the window. I believe the ICCCM et al. may go some way to solving this problem. -- RJ Noble, Victoria University of Wellington, New Zealand kjx@comp.vuw.ac.nz ...!uunet!vuwcomp!kjx
duanev@kauai.ACA.MCC.COM (Duane Voth) (10/18/89)
In article <2779@ninja.Solbourne.COM>, toml@Solbourne.COM (Tom LaStrange) writes: > What I almost need is a standard -ignore "some string here" option that > all clients will ignore. This string could then be looked for in the > WM_COMMAND property when the window manager goes to reparent the > window. The window manager could parse the string for any options to > restore the client to the state from which it was saved. Hear hear! I want to add weight to this idea. I think we are going to find several places where a window manager is going to need configuration information specific to each *instance* of an application. -- --- duane voth duanev@mcc.com ---- ALL systems are arbitrary! --- effectiveness is the measure of Truth --