Problem executing a script twice

ROOT
1

Hello all,
I have a problem when I execute a specific Root script twice.
In order to illustrate it, I have made a very minimalistic piece of code which reproduce the problem:

[code]#include “TGraph.h”
#include

void vectors()
{
Int_t Npoints = 10;
vector<Float_t> x(Npoints);
vector<Float_t> y(Npoints);

TGraph *g = new TGraph();
for(Int_t i=0;i<Npoints;i++) {
g->Set(i+1);
x[i] = i+1;
y[i] = x[i]*x[i];
g->SetPoint(i,x[i],y[i]);
}

g->Draw(“APC”);
}
[/code]

If one runs it once from CINT it works like a charm, plotting this nice parabola, but if you execute it again, it crashes miserably.

Any clues?
Thanks in advance.

2

[code]#include “TInterpreter.h”
#include “TGraph.h”

#if !defined(CINT)
#include
#endif /* !defined(CINT) */

void vectors()
{
#if defined(CINT)
if ( ! (gInterpreter->IsLoaded(“vector”)) ) {
#if 1 /* 0 or 1 /
gInterpreter->ProcessLine("#include ");
#else / 0 or 1 /
gInterpreter->Load(“vector”);
#endif / 0 or 1 /
}
#endif / defined(CINT) */

Int_t Npoints = 10;
vector<Float_t> x(Npoints);
vector<Float_t> y(Npoints);

TGraph *g = new TGraph();
for(Int_t i=0;i<Npoints;i++) {
g->Set(i+1);
x[i] = i+1;
y[i] = x[i]*x[i];
g->SetPoint(i,x[i],y[i]);
}

g->Draw(“APC”);
}[/code]
BTW. You can execute “.x vectors.cxx” many times without any problems. There is still a problem (i.e. a “segmentation violation”) if one first does “.x vectors.cxx+” (one plus) and then “.x vectors.cxx++” (two pluses). No problems if one first does “.x vectors.cxx++” (two pluses) - then the next “.x vectors.cxx+” (one plus) and/or “.x vectors.cxx++” (two pluses) are fine.

[code]root [0] .x vectors.cxx+
Info in TCanvas::MakeDefCanvas: created default TCanvas with name c1
root [1] .x vectors.cxx++
Info in : unmodified script has already been compiled and loaded
Info in : it will be regenerated and reloaded!
Info in TUnixSystem::ACLiC: creating shared library /…/./vectors_cxx.so

*** Break *** segmentation violation

===========================================================
There was a crash (#7 0xb72a616d in SigHandler(ESignals) ()).
This is the entire stack trace of all threads:

#0 0xb77e1430 in __kernel_vsyscall ()
#1 0xb662f7d3 in __waitpid_nocancel ()
at …/sysdeps/unix/syscall-template.S:82
#2 0xb65d0de3 in do_system (line=)
at …/sysdeps/posix/system.c:149
#3 0xb670027d in system (
line=0x8416730 “/…/v5-30-00-patches/etc/gdb-backtrace.sh 13814 1>&2”) at pt-system.c:29
#4 0xb72a016d in TUnixSystem::Exec(char const*) ()
from /…/v5-30-00-patches/lib/libCore.so.5.30
…
#38 0xb687ca8b in TRint::Run(bool) ()
from /…/v5-30-00-patches/lib/libRint.so.5.30
#39 0x08048edf in main ()

The lines below might hint at the cause of the crash.
If they do not help you then please submit a bug report at
http://root.cern.ch/bugs. Please post the ENTIRE stack trace
from above as an attachment in addition to anything else
that might help us fixing this issue.

#12 0xb71a1a0a in TApplication::IsCmdThread() ()
from /…/v5-30-00-patches/lib/libCore.so.5.30

Root > !!!Dictionary position not recovered because G__unloadfile() is used in a macro!!!

*** Break *** segmentation violation

===========================================================
There was a crash (#7 0xb72a616d in SigHandler(ESignals) ()).
This is the entire stack trace of all threads:

#0 0xb77e1430 in __kernel_vsyscall ()
#1 0xb662f7d3 in __waitpid_nocancel ()
at …/sysdeps/unix/syscall-template.S:82
#2 0xb65d0de3 in do_system (line=)
at …/sysdeps/posix/system.c:149
#3 0xb670027d in system (
line=0x84d7358 “/…/v5-30-00-patches/etc/gdb-backtrace.sh 13814 1>&2”) at pt-system.c:29
#4 0xb72a016d in TUnixSystem::Exec(char const*) ()
from /…/v5-30-00-patches/lib/libCore.so.5.30
…[/code]

3

Dear Pepe Le Pew,
If I use either “.x vectors.C++” or “.x vectors.C+”, the script works in its original version, in any possible combination of “+” and “++”. There is no need to add these “CINT” lines you proposed.
The problem I have is when not using the “+” : “.x vectors.C”.

Nevertheless, if we slightly modify the script to get the number of points in the parabola as an argument, it doesn’t work anymore after the first execution. Try this code, please:

[code]#include “TGraph.h”
#include

void vectors(Int_t Npoints = 10)
{
vector<Float_t> x(Npoints);
vector<Float_t> y(Npoints);

TGraph *g = new TGraph();
for(Int_t i=0;i<Npoints;i++) {
g->Set(i+1);
x[i] = i+1;
y[i] = x[i]*x[i];
g->SetPoint(i,x[i],y[i]);
}

g->Draw(“APC”);
}
[/code]

Then execute it modifying the argument:
".x vectors.C(20)+" will crash immediately.
".x vectors.C(20)++" will work the first time, but it won’t a second.

Thanks!

4

.x vectors.C+(20)
.x vectors.C++(20)

Use my proposed modifications (i.e. my “brutal fix”) in order to “overcome” the interpreted code problems (i.e. when you try “.x vectors.C(20)”).

[code]#include “TInterpreter.h”
#include “TGraph.h”

#if !defined(CINT)
#include
#endif /* !defined(CINT) */

void vectors(Int_t Npoints = 10)
{
#if defined(CINT)
if ( ! (gInterpreter->IsLoaded(“vector”)) ) {
#if 1 /* 0 or 1 /
gInterpreter->ProcessLine("#include ");
#else / 0 or 1 /
gInterpreter->Load(“vector”);
#endif / 0 or 1 /
}
#endif / defined(CINT) */

vector<Float_t> x(Npoints);
vector<Float_t> y(Npoints);

TGraph *g = new TGraph();
for(Int_t i=0;i<Npoints;i++) {
g->Set(i+1);
x[i] = i+1;
y[i] = x[i]*x[i];
g->SetPoint(i,x[i],y[i]);
}

g->Draw(“APC”);
}[/code]

5

Oh! Sorry, I did a mistake. When running the script with the arguments and the pluses the syntax is like you wrote: “.x vectors.C++(20)”, and not “.x vectors.C(20)++” as I tried…
In such a way the script works without any modification. Your “brutal fix” is not needed.
At least in ROOT 5.32.
Thanks a lot.

6

Hi Pepe and others,
This issue is still causing me a lot of troubles.
It is true that in this simple example, re-executions of the script inside the same CINT session are working if one uses the “+” and/or the “++”, but it doesn’t with other scripts more complicated.

There is something here which I don’t understand.
How is possible that a program which works with no problem once gets broken when it is called multiple times?
Everything seems to be due to multiple calls to #include .
In my other scripts I got warning messages of this type, when I re-execute them:

Warning: template pair duplicate definition /Users/delaossa/Programs/root_v5.32.00/cint/cint/stl/_pair.h:31: Warning: template reverse_iterator duplicate definition /Users/delaossa/Programs/root_v5.32.00/cint/cint/stl/_iterator.h:269: Warning: template allocator duplicate definition /Users/delaossa/Programs/root_v5.32.00/cint/cint/stl/defalloc.h:134: Warning: template vector duplicate definition /Users/delaossa/Programs/root_v5.32.00/cint/cint/stl/_vector.h:44: ...

Besides, If I compile a program which does multiple calls to these macros in a loop of this type:

// Time looper for(Int_t i=iStart; i<iEnd+1; i+=iStep) { // Run macros for plots TString command; command = Form(".x PlotPhase1D.C(%i)",time); gROOT->ProcessLine(command); }
I get segmentation faults although the script PlotPhase1D.C is working good in CINT (even with multiple calls) and the program itself works when it only loops once. Crazy…

Cheers!

7

Try my “brutal fix” (I implemented it precisely for the interpreted code).

8

.x vector.cxx(20)
vector(30)[/code]without problem.

Cheers,
Philippe.

PS. The original problem seems to be fixed in v5.34 (it ‘just’ prints harmless warning message).