Re: Pythia6 problem on Mac OS X Tiger

Hello,

thanks, that helped and it's a good reference on why that error occurs. I updated TPythia6.cxx to CVS HEAD and after recompilation it now works. Thanks everybody.

Mario

On Dec 22, 2005, at 9:48 AM, Dmitry Naumov wrote:

> Hi Mario,
> I encountered this problem recently. It happened to me because of
> the following:
> 1. the new compiler g77 (I have 3.4.5 you have 3.4.4) puts NO
> underscore at the end of fortran common_block
> and puts ONE underscore at the end of fortran routines.
> 2. The root "configure" scripts tries to determine of how many
> underscores are in the libPythia.so for
> pythia6_common_block_address and respecting this it produces a
> special compilation option (flag) for the
> Pythia (like: -DPYTHIA6_DOUBLE_UNDERSCORE, -
> DPYTHIA6_SINGLE_UNDERSCORE, -DPYTHIA6_NO_UNDERSCORE)
> This scripts assumes that the same rule is applied also for
> fortran routines. It works with previous compilers.
> However new g77 seems to change its behaviour and now they
> assign no underscore for common blocks and one (or two) underscores
> for routines. That is why to my mind the scripts produces a
> wrong option.
> 3. As a result TPythia6.cxx can not find a proper interface to
> these two functions:
> tpythia6_close_fortran_file
> tpythia6_open_fortran_file
> 4. There are two solutions at least:
> o) add a new option into configure script (one for common
> blocks, second for routines). This is an easy thing to do (I can do
> it if needed)
> o) just put by hand (or take from CVS head as suggested by Rene)
> the following patch into TPythia6.cxx:
> # ifdef PYTHIA6_DOUBLE_UNDERSCORE
> # define tpythia6_open_fortran_file tpythia6_open_fortran_file__
> # define tpythia6_close_fortran_file tpythia6_close_fortran_file__
> # define pythia6_common_address pythia6_common_block_address__
> # elif PYTHIA6_SINGLE_UNDERSCORE
> # define tpythia6_open_fortran_file tpythia6_open_fortran_file_
> # define tpythia6_close_fortran_file tpythia6_close_fortran_file_
> # define pythia6_common_address pythia6_common_block_address_
> # else
> # define pythia6_common_address pythia6_common_address
> # define tpythia6_open_fortran_file tpythia6_open_fortran_file_
> # define tpythia6_close_fortran_file tpythia6_close_fortran_file_
> # endif
>
> lines after else were missing for this configuration. This is not
> really a nice solution but it can solve your problem temporarly.
> Hope this can help you,
> Regards, Dmitry
>
> Mario Kadastik wrote:
>
>> maku:/Applications/root/lib toruonu$ nm libPythia6.dylib |grep
>> tpythia
>> 00001170 T _tpythia6_close_fortran_file_
>> 000010f4 T _tpythia6_open_fortran_file_
>> maku:/Applications/root/lib toruonu$ nm libPythia6.so |grep tpythia
>> 00001190 T _tpythia6_close_fortran_file_
>> 00001114 T _tpythia6_open_fortran_file_
>>
>> so the relevant functions are there... Am I the idiot here or why
>> does g++ just not work?
>>
>> Mario
>>
>> On Dec 22, 2005, at 12:47 AM, Edmond Offermann wrote:
>>
>>> try "nm"
>>>
>>> --- Mario Kadastik <mario.kadastik_at_cern.ch> wrote:
>>>
>>>> Hello Rene,
>>>>
>>>> Well the files are in tarball, but you have to compile them to make
>>>> libPythia6.so and libPythia6.dylib. That's exactly what I did
>>>> and it
>>>>
>>>> also works just fine in CINT however it doesn't work when I try to
>>>> compile it.
>>>>
>>>> The versions I use are:
>>>> maku:~/Downloads/pythia6 toruonu$ g77 --version
>>>> GNU Fortran (GCC) 3.4.4
>>>> Copyright (C) 2004 Free Software Foundation, Inc.
>>>>
>>>> GNU Fortran comes with NO WARRANTY, to the extent permitted by law.
>>>> You may redistribute copies of GNU Fortran
>>>> under the terms of the GNU General Public License.
>>>> For more information about these matters, see the file named
>>>> COPYING
>>>> or type the command `info -f g77 Copying'.
>>>> maku:~/Downloads/pythia6 toruonu$ gcc --version
>>>> powerpc-apple-darwin8-gcc-4.0.0 (GCC) 4.0.0 20041026 (Apple
>>>> Computer,
>>>>
>>>> Inc. build 4061)
>>>> Copyright (C) 2004 Free Software Foundation, Inc.
>>>> This is free software; see the source for copying conditions.
>>>> There
>>>>
>>>> is NO
>>>> warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR
>>>> PURPOSE.
>>>>
>>>> that's so as g77 latest release is for 3.4.4 after that there is
>>>> only
>>>>
>>>> fortran f95 and using that will not compile pythia as it fails with
>>>> missing symbols itime and idate ...
>>>>
>>>> And as I don't use gcc-4.0.3 then what else could be the problem? I
>>>> tried reorganizing the library reference on the command line hoping
>>>> it would cure the unresolved symbol, but no... is there a way to
>>>> check which symbols are in which libraries in Mac OS X so that I
>>>> could check if they are in the libraries?
>>>>
>>>> Mario
>>>>
>>>> On Dec 21, 2005, at 8:20 PM, Rene Brun wrote:
>>>>
>>>>> Mario,
>>>>>
>>>>> I do not understand your problem. The files that you quote as
>>>>> missing when linking are provided in the tar file.
>>>>> May be you are hitting the same problem that was reported recently
>>>>
>>>>
>>>>> by Dmitry Naumov when compiling with the latest gcc4.0.3 compiler
>>>>> (something to do with underscores).
>>>>> If indeed you use this compiler, you can start from the CVS head
>>>>> where this problem has been fixed. You can take pythia6/src/
>>>>> TPythia6.cxx
>>>>> only.
>>>>>
>>>>> Let me know
>>>>>
>>>>> Rene Brun
>>>>>
>>>>> On Wed, 21 Dec 2005, Mario Kadastik wrote:
>>>>>
>>>>>> Rene Brun wrote:
>>>>>>
>>>>>>> Mario,
>>>>>>> Look at http://root.cern.ch/root/Install.html
>>>>>>> and section "Pythia generator:
>>>>>>> Take the tar file pythia6.tar.gz. It contains all the necessary
>>>>>>> files
>>>>>>> to compile on all machines.
>>>>>>
>>>>>>
>>>>>> that's exactly where I took it from :) and that's the
>>>>>> makePythia.macosx script I'm running in my previous e-mail and as
>>>>>
>>>>
>>>>>> you can see it compiles nicely. I can also run my code in root,
>>>>>> but I can't compile it due to the linking errors I wrote about.
>>>>>> But I would like to compile as I have a lot of datasets against
>>>>>> which I'd like to run my code and running it from command line
>>>>>> using arguments is a lot more convenient.
>>>>>>
>>>>>> Mario
>>>>>>
>>>>
>>>>
>>>
Received on Thu Dec 22 2005 - 12:19:27 MET