[Clas_offline] user_ana on centos62 with svn/scons

[KiJun Park]

Dear Silvia,

It seems to me that your problem is related with link CERNLIB.
Have you tried to link CERNLIB with "/u/site/cernlib/x86_64_rhel6.old/2005/" ?

Sincerely,
Kijun



----- Original Message -----
From: "Silvia Niccolai" <silvia at jlab.org>
To: "Evan Phelps" <evan.phelps at gmail.com>
Cc: "KiJun Park" <parkkj at jlab.org>, "clas offline" <clas_offline at jlab.org>
Sent: Monday, April 29, 2013 5:15:39 AM
Subject: Re: [Clas_offline] user_ana on centos62 with svn/scons

Hello Evan,
I just modified the file recsis_proc.tcl according to your suggestion, 
recompiled the whole reconstruction/ packages, but I still get the same 
problem with user_ana, it remains stuck at the prompt. Just to be more 
precise, here are the last output lines I get:

***************************************************************
Defining to <paw> procedure, to launch a PAW session type: paw

Defining to <ced> procedure, to launch a ced session type: ced


RS> RZOPEN. record length:  8191 > maximum safe value (8191 words).
  RZOPEN. You may have problems transferring your file to other systems
          or writing it to tape.
bnames: 1000
RS>
***************************************************************

To be compared with what I get on ifarml64, with an old user_ana 
executable (same inputfile, same tcl):

**************************************************************
Defining to <paw> procedure, to launch a PAW session type: paw

Defining to <ced> procedure, to launch a ced session type: ced

recsis_proc I: sending FPACK CMD: OPEN CLASINPUT UNIT=01 FILE="inputfile" 
READ RECL=32768 STATUS=OLD

recsis_proc I: sending FPACK CMD: OPEN PROC1 UNIT=2 FILE="outfile1" WRITE 
RECL=32768 SPLITMB=2047 STATUS=NEW

...and then it runs fine...
****************************************************************

Did it instead work for you? Did you produce an actual cooked file as 
output? If yes, could you please let me know your scons command line?
Did you compile it at 32 or 64 bits? I tried first only with a bare 
"scons" command. The produced executable is the one giving me the above 
problem. I tried forcing 32 bits compilation, but got library/environment 
problems (I currently work with an environment provided to me by Johann).

Any tips would be greatly appreciated.

Best regards,
Silvia



On Fri, 26 Apr 2013, Evan Phelps wrote:

> Kijun,
>
> I don't know if this is the same issue, but I'll pass on the relevant
> part of a message I sent last December regarding getting dumped back
> to the recsis prompt without error (see below).  The culprit, in my
> case, was the tcl file, but after getting past that, I ran into
> additional library incompatibilities that I have not revisited.
>
> *****************
>
> [...]  I noticed that the tcl procedure "inputfile" calls "fpack," and
> that procedure does not complete its execution.  It seems to stop and
> return to recsis on the following line:
>
> global rchar_control ifpack \
>    fpk_cmd(1) fpk_cmd(2) fpk_cmd(3) fpk_cmd(4) fpk_cmd(5) \
>    fpk_cmd(6) fpk_cmd(7) fpk_cmd(8) fpk_cmd(9) fpk_cmd(10)
>
> I'm not sure why this works in previous, non-CentOS62, environments,
> since the TCL manual says that array elements cannot be referenced
> globally.  It seems to work after correcting that. Here's the diff:
>
> [ephelps at ifarm1101 sim]$ diff
> /group/clas/builds/2012.12.20_10.42/trunk/reconstruction/recsis/recsis_proc.tcl
> ./recsis_proc.tcl
> 282,285c282
> <     global rchar_control ifpack \
> <     fpk_cmd(1) fpk_cmd(2) fpk_cmd(3) fpk_cmd(4) fpk_cmd(5) \
> <     fpk_cmd(6) fpk_cmd(7) fpk_cmd(8) fpk_cmd(9) fpk_cmd(10)
> <
> ---
>>     global rchar_control ifpack fpk_cmd
> 343c340
> <     global rchar_control fpk_cmd(1)
> ---
>>     global rchar_control fpk_cmd
>
> On Fri, Apr 26, 2013 at 9:49 AM, KiJun Park <parkkj at jlab.org> wrote:
>> Hi,
>>
>> I don't know whether this can help to solve your problem but ...
>> Last week, Jlab CC installed "compat-libgfortran-41.i686", so maybe correction of your path to new-link could be an issue.
>>
>> Sincerely,
>> Kijun
>>
>>
>>
>>
>> ----- Original Message -----
>> From: "Silvia Niccolai" <silvia at jlab.org>
>> To: "Paul Mattione" <pmatt at jlab.org>
>> Cc: "clas offline" <clas_offline at jlab.org>
>> Sent: Monday, April 22, 2013 2:34:15 PM
>> Subject: Re: [Clas_offline] user_ana on centos62 with svn/scons
>>
>> Hello Paul,
>> I managed to build recsis and to have a user_ana executable on ifarm1102, but I had your same problem as you when launching it: it pauses and gives the prompt and seems like it hasn't found the inputfile, even if all is well defined in the tcl.
>> Any input/suggestion would be greatly appreciated!!
>> Regards,
>> Silvia
>>
>>
>> Sent from my iPhone
>>
>> On 22/apr/2013, at 20:28, Paul Mattione <pmatt at jlab.org> wrote:
>>
>>> Has anyone gotten an svn/scons build of user_ana to work (meaning: both compiled & run successfully with meaningful output) on simulated data on ifarm1102 (either a 32- or 64-bit build is fine)?  I tried searching the CLAS6 offline software wiki page and did not see any documentation/details on the currently existing builds.
>>>
>>> I have tried many different svn/scons builds and they do not work on my (old, previously working) MC tcl file, whereas the current ifarm1102 g14 make build works just fine on it.  The farthest I get on the svn/scons builds is that the cooking pauses during reading the tcl file, giving the user a prompt and expecting more input for some reason.  In addition to my own (g13), the attempted builds include (all within /group/clas/builds/):
>>>
>>> test3
>>> 64bit (is this an ifarm1101 build?  It is unclear.)
>>> 2012.12.20_10.42 (ifarm1101 or ifarm1102?)
>>> centos62 (In addition, neither environment file has the correct $PREFIX assignment)
>>>
>>> Am I using the wrong build(s)?  If user_ana for these builds does indeed work, can someone send me a proven, working tcl file and full environment for them so I can try to diagnose what might be different between the make & scons builds of our software?
>>>
>>> I realize that everyone is busy with CLAS12, but CLAS6 software still needs to be supported.  Why hasn't fsgen been integrated into svn/scons yet?!?!?  I asked for this many months ago and it still isn't available.
>>>
>>> Thanks.
>>>
>>> - Paul
>>>
>>> _______________________________________________
>>> Clas_offline mailing list
>>> Clas_offline at jlab.org
>>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>>
>> _______________________________________________
>> Clas_offline mailing list
>> Clas_offline at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>>
>> --
>> ---------------------------------------******
>> Dr. Kijun Park
>> Thomas Jefferson National Accelerator Facility
>> Suit 5 Room 12_1/A102
>> 12000 Jefferson Ave.
>> Newport News VA 23606
>> Phone 757-269-6989
>>
>> _______________________________________________
>> Clas_offline mailing list
>> Clas_offline at jlab.org
>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>

-- 
---------------------------------------******
Dr. Kijun Park
Thomas Jefferson National Accelerator Facility
Suit 5 Room 12_1/A102
12000 Jefferson Ave.
Newport News VA 23606
Phone 757-269-6989