Hi Rene and Federico,
I see a very similar (probably the same) problem on RH6.1/egcs-1.1.2. I
wrote a _standalone_ program which reads data from a TNtuple, fills a
TGraphErrors, draws it (so far everything works) but when I fit it with a
TF1 (a very simple one, a single exponential), it crashes right after the
fitting (fit results are printed correctly) and the fitted function never
appears on the canvas. gdb says:
Program received signal SIGSEGV, Segmentation fault.
0x408407c6 in TGraph::PaintFit ()
from /home/juhaszbe/install/root/new/lib/libGraf.so
(gdb) bt
#0 0x408407c6 in TGraph::PaintFit ()
from /home/juhaszbe/install/root/new/lib/libGraf.so
#1 0x40841646 in TGraph::PaintGraph ()
from /home/juhaszbe/install/root/new/lib/libGraf.so
#2 0x40840125 in TGraph::Paint ()
from /home/juhaszbe/install/root/new/lib/libGraf.so
#3 0x4084c170 in TGraphErrors::Paint ()
from /home/juhaszbe/install/root/new/lib/libGraf.so
#4 0x40a338cc in TPad::PaintModified ()
from /home/juhaszbe/install/root/new/lib/libGpad.so
#5 0x40a33876 in TPad::PaintModified ()
from /home/juhaszbe/install/root/new/lib/libGpad.so
#6 0x40a1c007 in TCanvas::Update ()
from /home/juhaszbe/install/root/new/lib/libGpad.so
#7 0x804a5a4 in ?? ()
#8 0x411201eb in __libc_start_main (main=0x80496d8, argc=4,
argv=0xbffffaf4,
init=0x804935c, fini=0x804b384, rtld_fini=0x4000a610 <_dl_fini>,
stack_end=0xbffffaec) at ../sysdeps/generic/libc-start.c:90
The .so files (version 3.03/07) are directly from the ROOT website. Same
crash happens if I try to fit using the FitPanel.
Regards,
Bertalan
On Wed, 14 Aug 2002, Rene Brun wrote:
> Hi Federico,
>
> I cannot reproduce your problem on my system (RH6.1,gcc295.2).
> I have no access to a RH7.2/gcc2.96 before tomorrow.
> Here is what I get when running your script
> 911 lines counted
> Device Name: bwsh51995.rot
> Scan Direction: OUT
> 905 data points found
> Peak= 1886
> XLOW 10426.3 XHIGH 12159.7
> par0=11137.4 par1=781.138
> **********
> ** 1 **SET ERR 1
> **********
> drawing
> drawn
>
> RESULTS:
> RMS Data = 781.137545
> FWHM = 866.724589
> Sigma Fit1 = 781.131624
> Sigma Fit2 = 781.137545
> file written
> done
>
> and the corresponding c.gif file in attachement.
> Let me know if both output look reasonable.
> Note that in your code, you define twice ifstream finput!
>
> Could somebody running RH7.2 with either gcc2.95, 2.96 or 3.0,1
> run this test to see if you also get a crash.
> Could you run under gdb and tell me the traceback when the program
> crashes?
>
> Rene Brun
>
> On Wed, 14 Aug 2002, Federico Roncarolo wrote:
>
> > Hello,
> >
> > I'm using version 3.03/07 on linux/redhat 7.2.
> >
> > I'm loading my code with .L and then calling main(argument).
> >
> > I'm getting data from an ascii file, filling graphs and fitting the
> > graphs.
> > Later when I just draw the graph I expect to see the graph with the
> > fitting function, but root crashes, it displays an empty canvas and gives
> > a
> >
> > *** Break *** segmentation violation.
> >
> >
> > I fit with a user-defined function with the options "QRME+".
> > If I add the options "NO", when drawing the graph everything works, but of
> > course I get only the graph and not the function.
> >
> > The segmentation violation comes at the end of code and not just after
> > drawing.
> >
> > Please find attached the file.C and the data file I give as argument of
> > the main function.
> > The line where I fit is number 252.
> >
> > I changed from RH7.1 to RH7.2 ten days ago and re-installed ROOT from the
> > file
> >
> > root_v3.03.07.Linux.RH7.2.gcc296.tar.gz
> >
> > I test few small codes and they look to work... any idea about what is
> > wrong?
> >
> > Thanks a lot,
> > Federico
> >
>
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:51:04 MET