#2026 menu bugs with huge display (xinerama)

obsolete: 8.3.2

Solaris 8, openwin X11 server, CDE window manager
I'm running X in "Xinerama" mode with one logical
across two physical displays. To X clients, it appears
as if there is one single 2560x1024 display.

Both of the problems below occur when a Tk window is
to the right half of the screen (into the second
monitor), but it
doesn't happen exactly at the same spot each time.
I can get a window fully into the right monitor, plus
20 or 30 pixels without problems, but then if I move it
farther to
the right, the problems start.

First, menus (at least within a menubar) become very
and unresponsive. They don't respond at all to
menu selection. Once started, this problem persists
of where the window is posistioned.

Second when submenus of a menu are posted, they appear
the correct y, but far to the left, in the first
monitor. This problem
eventually goes away if the window is moved to the left
but also not at a consistent x-position.

I checked winfo: x, y, rootx, rooty, screenwidth,
pointerx, and pointery all return values that you'd
expect for the
main window regardless of where the window is. However
I post a menu with ".mb1.menu1 post 1800 120", and the
immediately do "winfo rootx .mb1.menu1", it returns
520, which is
1800-1280, and 1280 is half the screen width, one
monitor width.
The menu still appears in the right half of the screen
though. If I
then call ".mb1.menu1 postcascade 2", the sub menu
winds up
(not surprisingly) at an x of 639, both visibly and as
by winfo rootx. (winfo x reports the same as winfo
rootx in these
two cases also.)

dtwm uses a "rooms" virtual window paradigm, so vrootx
vrooty should always be zero, and they are reported as
vrootwidth and vrootheight are also reported correctly


    • status: open --> open-invalid
  • Logged In: YES

    This is a Tk bug, not a Tcl one.

  • Logged In: NO

    OK, this is extremely unhelpful. Has this been
    transferred to a Tk bug database? If not, where do I go to
    do that?
    Why does the Tcl/Tk home page direct me to this bug database
    for "Tcl/Tk core" bugs, if I can't report Tk bugs here?

  • Logged In: YES

    I know it is unhelpful, but SF (in their infinite wisdom)
    removed the feature that allowed people with appropriate
    permissions to move reports between projects. Annoying but
    the alternatives (move to some other hosting scheme for the
    core) have other costs; SF provides most of what is wanted,
    and for free.

    The project you want is the 'tktoolkit' project. (If I'd
    been an expert on xinerama I might have given more
    information about what to do, but alas, I know nothing.)

    I won't close this bug until I've seen it transferred, but
    I'd prefer someone else did it (especially if they log in
    first) so that they get to watch what's going on instead of
    having to poll the bug DB... :^/

  • Thomas A. Fine
    Thomas A. Fine

    Logged In: YES

    You can close this. I submitted a new bug to the Tk bug
    with identical text to this one.

    There is a (very easy to miss) link to the Tk bug database
    from the
    Tcl bug database, which I found when I looked again.

    I've also created an account for myself. You know the for
    new bug reports doesn't have anywhere to enter an email
    address if
    you choose not to create an account, which is what I did the
    first time.

    Things should just work. Grumble, grumble, grumble.

    • assigned_to: nobody --> dkf
    • status: open-invalid --> closed-invalid
  • Logged In: YES

    Ah, you want "Do what I *really* mean!" software? Join the
    (long) queue... ;^)
    (I believe talking to Don Porter - dgp@users.sf.net - is the
    way forward when it comes to improving the website; I'm not
    involved there...)