I recently compiled ROOT on a Sun CC5 system using the old iostream (thus with switch -library=iostream,no%Cstd) and after some changes (now in CVS) this resulted in a working version. However, when trying without this iostream option and using the sunstrm.cxx I did have problems with sunstrm.cxx. Eddy, try the version in CVS which should work on Sun CC5 (with compatability iostream). George, if you could get the system to compile with the real iostream please mail me the mods. Cheers, Fons. On Wed, Nov 29, 2000 at 02:22:53PM -0500, George A. Heintzelman wrote: > > > /opt/SUNWspro/bin/CC -O -library=iostream,no%Cstd -o utils/src/rootcint_tmp \ > > utils/src/rootcint_tmp.o cint/src/auxu.o cint/src/cast.o > > cint/src/debug.o cint/src/decl.o cint/src/disp.o cint/src/dump.o cint/src/end.o > > cint/src/error.o cint/src/expr.o cint/src/fread.o cint/src/func.o > > cint/src/g__cfunc.o cint/src/gcoll.o cint/src/global1.o cint/src/global2.o > > cint/src/ifunc.o cint/src/inherit.o cint/src/init.o cint/src/input.o > > cint/src/intrpt.o cint/src/loadfile_tmp.o cint/src/macro.o cint/src/malloc.o > > cint/src/memtest.o cint/src/new.o cint/src/newlink.o cint/src/newsos.o > > cint/src/oldlink.o cint/src/opr.o cint/src/parse.o cint/src/pause.o > > cint/src/pcode.o cint/src/pragma.o cint/src/quote.o cint/src/scrupto.o > > cint/src/shl.o cint/src/sizeof.o cint/src/stdstrct.o cint/src/struct.o > > cint/src/stub.o cint/src/tmplt.o cint/src/typedef.o cint/src/val2a.o > > cint/src/value.o cint/src/var.o cint/main/G__setup.o cint/src/Api.o > > cint/src/Apiif.o cint/src/BaseCls.o cint/src/CallFunc.o cint/src/Class.o > > cint/src/DataMbr.o cint/src/Method.o cint/src/MethodAr.o cint/src/Token.o > > cint/src/Type.o cint/src/Typedf.o cint/src/sunstrm.o -lm -ltermcap -ldl > > Undefined first referenced > > symbol in file > > std::basic_string<char,std::char_traits<char>,std::allocator<char> > > >std::basic_ostringstream<char,std::char_traits<char>,std::allocator<char> > > >::str()const cint/src/sunstrm.o > > > > This morning I saw a mail flashing by from George/Masaharu concerning > > sunstrm.cxx. Me being to trigger-happy deleted it. Could the solution > > to my compilation problem have been contained in that email ?? > > I don't think so, but there are a lot of issues with the Sun stuff that > is in the current CVS. When I compile on the sun, I don't use > -library=iostream,no%Cstd, which is unnecessary and I think is what is > causing it not to pick up this C++ standard library function. Instead, > what I use is this: > > CXXFLAGS = -KPIC -I/usr/openwin/include -D__SunOS_5_6 \ > -DR__ANSISTREAM $(EXTRA_CXXFLAGS) -ptr$(ROOTSYS) > $(THREAD) > CFLAGS = -KPIC -I/usr/openwin/include $(EXTRA_CFLAGS) $(THREAD) > CINTCXXFLAGS = -KPIC -DG__REGEXP1 -DG__UNIX -DG__OSFDLL > $(EXTRA_CXXFLAGS) \ > -DG__SHAREDLIB -DG__ROOT -DG__REDIRECTIO -ptr$(ROOTSYS) > \ > $(THREAD) > CINTCFLAGS = -KPIC -DG__REGEXP1 -DG__UNIX -DG__OSFDLL > $(EXTRA_CFLAGS) \ > -DG__SHAREDLIB -DG__ROOT -DG__REDIRECTIO $(THREAD) > > Unfortunately, the sun system I normally work on has been undergoing > system work and I can't verify why all these are needed for you. I do > know that the -ptr$(ROOTSYS) is so that it uses a common cache (which I > think is mostly for template stuff, but might have other purposes) for > the whole compile. Threading does seem to work as well, with THREAD = > -mt, although I don't claim to have tested it more than superficially. > The sun libraries are now fairly close to ANSI compliant (bearing in > mind that CC 5.0 doesn't support member templates), so R__ANSISTREAM is > right. > > I'm not at all sure what the status of the ROOT in CVS would be under > these flags; the last I compiled on SUN was a somewhat modified version > of 2.25/02. > > George Heintzelman > gah@bnl.gov > > > -- Org: CERN, European Laboratory for Particle Physics. Mail: 1211 Geneve 23, Switzerland E-Mail: Fons.Rademakers@cern.ch Phone: +41 22 7679248 WWW: http://root.cern.ch/~rdm/ Fax: +41 22 7677910
This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:38 MET