#1417 With the 'qt' terminal, the initial window size is incorrect

[Vincent Lefevre]

With the gnuplot-qt 4.6.5-4 Debian package under Debian/unstable and the 'qt' terminal (which is now the default), the initial window size is incorrect. This can be seen with fvwm's active placement (because the frame of the initial window is displayed until the window is placed by the user).

The consequences are the following if the window manager places the window depending on its initial size:

• The window can be partly off-screen.
• The window can cover other windows (in particular the one that is used to run gnuplot from a shell), even though there is enough empty space for it elsewhere.

I could reproduce both problems on my machine.

The testcase:

$ gnuplot -persist <<EOF
plot '-' using 1:2 t '' with line
0 0
10 10
e
EOF

There is no such problem if one uses "set terminal wxt" (when it was working, i.e. with libwxgtk2.8-0) or "set terminal x11".

[Vincent Lefevre]

I forgot to give the reference of the Debian bug:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=668956

[Ethan Merritt]

I don't understand the report. You have not specified a size in the "set term" command, so on what basis do you conclude it is the "wrong" size?

If you say "set term qt size 500,300" or whatever, do you not get that size?

[Vincent Lefevre]

gnuplot opens the window with some size, then resizes it to its final size, leading to the observed problem.

If I write "set term qt size 500,300", the problem is the same: the initial size is wrong and the final size is 500×300.

In xtrace output, I see lots of "width=243 height=113". I wonder whether this is related to this problem.

[Vincent Lefevre]

Well, the initial size is approximately 243×113, so that these "width=243 height=113" are probably related to this problem.

[Ethan Merritt]

OK. You are probably seeing the same problem that is being tracked on this ticket:

https://sourceforge.net/p/gnuplot/patches/661/

As you will see there, this problem is reproducible on some machines but not others. At least one of the other reporters is using Debian, for whatever that is worth. There is a patch there as well.

[Vincent Lefevre]

I don't know, the code has significantly changed. There's no !parent->isVisible() test in 4.6.5.

[Ethan Merritt]

Duplicate - see Bug 661