Tk Library Source Code

We have a tool that uses the BWidget MainFrame to display 
data on a number of panes and notebook pages. Part of the 
functionality is that it can save and restore the configuration 
of the various subdivisions within the frame. 
 
We found that if these were saved under Windows and then 
restored under Linux a segmentation fault resulted. 
 
Attached is the file mre.tcl which implements our tool, and 
the saved state file from Windows which causes the crash 
when loaded under Linux (Red Hat 7.3/8.0) 
 
Here's how to reproduce it: 
[jaspert@spirograph Run]$ wish 
% source mre.tcl 
% RunEnv::Create dummy 
.mre 
% open EmptyWindows.shf r 
file4 
% RunEnv::LoadViewFile file4 {600 400 0 0} 
Segmentation fault 
 
Also attached is a corresponding viewfile produced under 
Linux which does not cause the crash

Logged In: YES 
user_id=72656

I went ahead and fixed this for 8.4.  The code is actually an 
error - the first sash only exists when 2 panes exit.  This is in 
8.4.2.  You will get an error now from your original code 
sequence.

Logged In: YES 
user_id=72656

It's much easier on developers when problems like this are 
distilled down.  This can be repeated in 3 lines of code:

    panedwindow .p
    .p add [frame .f -width 20 -height 20 -bg pink]
    .p sash place 0 2 5

It's also a core bug, not a bwidgets bug.  Please reissue in 
the Tk tracker (unfortunately SF doesn't allow moving bugs 
between projects).

File Added - 43020: mre.tcl

Logged In: YES 
user_id=456242

OK I've edited mre.tcl and removed the references to other things 
that were lying around my system (I hope...) Also I found that after 
the frame is drawn by the RunEnv::Create dummy command, you 
can get the segfault directly with the command 
 
.mre.mainframe.frame.mainpw.mainDisplayPane.notebook.fpage/1.panedwindow 
sash place 0 2 5 
 
(but change that last 5 into an 8 and it accepts the command) 
Hopefully that makes it easier to trace

Logged In: YES 
user_id=72656

When trying with those files and your example, it complains 
about a missing image and nothing works ... perhaps rework 
without the images?

Logged In: YES 
user_id=456242

You should be able to reproduce the problem with these files. I 
didn't develop this code I am just trying to port the tool to Linux. If I 
have time I'll try and find a simpler procedure that causes the 
segfault.

Logged In: YES 
user_id=72656

The attached patch is not a self-contained example.  Could 
you narrow it to a smaller code snippet?

File Added - 42845: buginfo.tgz

Logged In: YES 
user_id=456242

Oops, seems my files didn't get uploaded, I'll try again