whizard is hosted by Hepforge, IPPP Durham
close Warning: Error with navigation contributor "BrowserModule"

Opened 15 years ago

Closed 15 years ago

Last modified 14 years ago

#134 closed defect (invalid)

gfortran is still not working on some linux machines

Reported by: Juergen Reuter Owned by: ALL
Priority: P1 Milestone:
Component: core Version: 2.0.0beta
Severity: blocker Keywords:
Cc:

Description

It seems that the old, old ticket (color flow problem!) has reopened again for gfortran, at least on 32 bit Linux. More details soon. This is the over-killer!

Change History (8)

comment:1 Changed 15 years ago by Juergen Reuter

I JUST CONFIRMED THIS WITH TRUNK REVISION 156003 ON LINUX 32BIT (DEBIAN EDGE). IF THIS IS NOT SOLVED, THERE IS NO SENSE IN CONTUNUING.

comment:2 Changed 15 years ago by Juergen Reuter

On demand of CS this is the anamnese of that error. Running the test_alpha_s script, we get:

libtool: link: (cd ".libs" && rm -f "processes.so.0" && ln -s "processes.so.0.0.0" "processes.so.0")
libtool: link: (cd ".libs" && rm -f "processes.so" && ln -s "processes.so.0.0.0" "processes.so")
libtool: link: ar cru .libs/processes.a  ugug.o processes.o
libtool: link: ranlib .libs/processes.a
libtool: link: ( cd ".libs" && rm -f "processes.la" && ln -s "../processes.la" "processes.la" )
| Loading process library 'processes'
| Process 'ugug': updating previous configuration
sqrts =    500.00000000000000
Beam data (collision):
 u   (mass = 0.0000000 GeV)
 gl  (mass = 0.0000000 GeV)
 sqrts = 500.000000000000 GeV
| Integrating process 'ugug'
[c(-3 ) c(1 -2) c(2 ) ]
*** WHIZARD bug:  Color flow mismatch (color loops should be closed)
WHIZARD run aborted.

In ancient times, this was a bug in some array functions in gfortran, does anyone remember!? Is anyone still alive?

comment:3 Changed 15 years ago by Juergen Reuter

Resolution: fixed
Status: newclosed

I'm awfully sorry, this was an ancient library from an LHAPDF installation which got intruded into the main system libraries of WHIZARD. With the new(er) library of gfortran the color bug is gone. Closing.

comment:4 Changed 15 years ago by ohl

Component: configurecore
Resolution: fixed
Status: closedreopened
Summary: gfortran is not working on 32 bit machinesgfortran is still not working on some linux machines

Sorry, this is not fixed:

with (it doesn't get much more recent)

ohl@wthp058:~whizard$ /archive/ohl/tools64/bin/gfortran --version
GNU Fortran (GCC) 4.5.0 20100117 (experimental)

I get on

Linux wthp058 2.6.26-2-amd64

the following

[  c(1 )  c(1 )]
*** WHIZARD bug:  Color flow mismatch (color loops should be closed)
WHIZARD run aborted.
FAIL: whizard_test_colors.sh

and

[c(-2 ) c(1 ) c(1 ) c(-1 )]
*** WHIZARD bug:  Color flow mismatch (color loops should be closed)
| There were no errors and 15 warning(s).
WHIZARD run aborted.
FAIL: whizard_test_modeltest.sh

comment:5 Changed 15 years ago by kilian

... puzzled.

No problem with

GNU Fortran (GCC) 4.5.0 20100111 (experimental)

on AMD64

Linux kagel 2.6.27.37-0.1-default #1 SMP 2009-10-15 14:56:58 +0200 x86_64 x86_64 x86_64 GNU/Linux

I'll update to the most recent version and check again.

BTW: We are all using default options, i.e., -g -O2, right?

comment:6 Changed 15 years ago by kilian

Update: with today's gfortran, no problem either.

comment:7 Changed 15 years ago by Christian Speckner

Resolution: invalid
Status: reopenedclosed

This is a runtime library mismatch between different versions of libgfortran.so.3. I have doublechecked on wthp058 using the same toolchain as Thorsten, and it works find if LD_LIBRARY_PATH is set correctly, and fails if not. If all else fails, try LD_PRELOAD to force the correct libraries. Closing this :)

comment:8 Changed 14 years ago by Juergen Reuter

Milestone: v2.0.0final

Milestone v2.0.0final deleted

Note: See TracTickets for help on using tickets.