Hi Remi, thanks for pointing out these issues. On Wed, 2004-06-09 at 23:36, Remi Mommsen wrote: > Hi, > > I compiled the current root version 4.00/06(a) on Mac OS X. I > encountered three problems. > > (1) On Mac OS X 10.3.4 the dynamic library lookup produces errors on > the first attempt. It succeeds on the 2nd attempt: > > root [0] TDatabasePDG pdg > dlopen error: dlcompat: dyld: /sw/bin/root.exe Undefined symbols: > _hepevt_ > > Load Error: Failed to load Dynamic link library /sw/lib/root/libEG.so > Error in <TCint::AutoLoadCallback>: failure loading library libEG.so > for class TDatabasePDG > *** Interpreter error recovered *** > root [1] TDatabasePDG pdg > root [2] TMCParticle part > dlopen error: dlcompat: dyld: /sw/bin/root.exe Undefined symbols: > __ZN10TGenerator11ShowMembersER16TMemberInspectorPc > __ZN10TGenerator12ExecuteEventEiii > __ZN10TGenerator12SetViewRangeEffffff > __ZN10TGenerator13SetViewRadiusEf > __ZN10TGenerator19DistancetoPrimitiveEi > Load Error: Failed to load Dynamic link library > /sw/lib/root/libEGPythia.so > Error in <TCint::AutoLoadCallback>: failure loading library > libEGPythia.so for class TMCParticle > *** Interpreter error recovered *** > root [3] TMCParticle part > > (The libraries exist at the specified locations.) > The problem is that Federico has us comment out the HEPEVT common in Hepevt.h, which of course causes this problem when libEG is loaded without pythia6 is loaded. Is there a better way and to have HEPEVT still defined in Hepevt.h and have it still shared with Pythia? > > > (2) I tried to build root 4.00/06(a) also on Mac OS X 10.2.8, but it > fails when trying to create the map file. It is reproducible on the > command line: > > $> bin/rlibmap -r etc/system.rootmap -l lib/libMatrix.dylib \ > -d -c matrix/inc/LinkDef.h > rlibmap: error unlocking output file > > Using '-o' instead of '-r' doesn't produce the error. > Yes, lockf() is new in 10.3, and even there is does not work. :-( > > (3) A minor issue. The build/unix/makelib.sh produces errors on Mac OS > X 10.2.8 because 'sw_vers' ignores the option '-productVersion'. Not > much harm is done because all if statements which fail due the wrong > formed conditional should fail anyhow. Here a patch which produces the > correct result for both OS X version: > > --- root/build/unix/makelib.sh Fri Mar 26 07:35:42 2004 > +++ root.new/build/unix/makelib.sh Wed Jun 9 17:19:04 2004 > @@ -75,7 +75,7 @@ > echo $LD $SOFLAGS$SONAME.$MAJOR.$MINOR $LDFLAGS -o > $LIB.$MAJOR.$MINOR `lorder $OBJS | tsort -q` $EXTRA $EXPLLNKCORE > $LD $SOFLAGS$SONAME.$MAJOR.$MINOR $LDFLAGS -o $LIB.$MAJOR.$MINOR > `lorder $OBJS | tsort -q` $EXTRA $EXPLLNKCORE > elif [ $PLATFORM = "macosx" ]; then > - macosx_minor=`sw_vers -productVersion | cut -d'.' -f2` > + macosx_minor=`sw_vers | sed -n > 's/ProductVersion:[[:blank:]]*[0-9]*.\([0-9]*\).[0-9]*/\1/p'` > # Look for a fink installation > FINKDIR=`which fink 2>&1 | sed -ne "s/\/bin\/fink//p"` > if [ $macosx_minor -ge 3 ]; then > Was not aware of the 10.2 incompatibility. Thanks. Cheers, Fons. -- Org: CERN, European Laboratory for Particle Physics. Mail: 1211 Geneve 23, Switzerland E-Mail: Fons.Rademakers@cern.ch Phone: +41 22 7679248 WWW: http://www.rademakers.org/fons/ Fax: +41 22 7679480
This archive was generated by hypermail 2b29 : Sun Jan 02 2005 - 05:50:08 MET