Re: [XMakemol-bugs] xmakemol-gl: export to xpm broken

[Matt Hodges]

>>>>> Wolfram Quester writes:

 > this is a bug that bothers me for quite some time now, but I was to
 > busy to report it. The problem occurs only with the xmakmol-gl
 > package, not without gl. If I want to export an image to xpm and
 > leave the export-dialog in front of the window I get the attached
 > result. It is gzipped since the first version bounced because it
 > was to large. One can see the molecule and the "shadow" of the
 > export window. At first sight I thought that in the "shadow" one
 > sees the content of the window behind xmakemol, but that's only
 > partially true. I the image I send, one sees only the titles of the
 > tabs in my gnome-terminal behind. It also occured that when I
 > switched between the tabs or from another workspace I got the
 > contant of this former tab/workspace, not the content of the
 > current one. I got this with version 5.14 from debian unstable as
 > well as with 5.15 (this one I started remotely on a machine runnung
 > AIX 5.1 and on my machine here). I use debian unstable on ppc
 > architecture.

This isn't really a bug, I think.  The comment in xmakemol.c explains
it:

    For some X11 displays, we can't draw on the canvas_pm due to
    crashes. Draw on the canvas window instead and use a backing store
    if available. See also gl_funcs.c

(That there used to be some crashes may have been the result of some
bug in XMakemol, but that is a different story.  Also, I don't want to
change this code if I can help it, since something might easily get
broken on systems where I cannot test it.)

Anyway, this gives the clue to work around the problem.  If you have a
backing store enabled, you shouldn't see the problem that you describe
above.  From my /etc/X11/XF86Config-4:

    Section "Device"
            Identifier          "Trident Cyberblade"
            Driver              "trident"
            Option              "BackingStore"
    EndSection

and from /var/log/XFree86.0.log:

    (**) TRIDENT(0): Option "BackingStore"
    (**) TRIDENT(0): Backing store enabled

(There may be solutions appropriate for X servers other than those
provided by XFree86.)

Does the above work for you on your Debian box?

Thanks,

Matt