Hi Paul, It is now clear that there is a serious interference between C++ and f90 on alpha. I think that Compaq should be contacted to get help. It is difficult to help from our side. Rene Brun Paul Nord wrote: > > Rene, > > To follow up on this old discussion. I'm now trying it with 2.23/12 and having > the same problem. I can get rid of the first of the error: > Unresolved: > __vtbl_3std9type_info__3std > by adding -nortti to ROOTCFLAGS. > > If I add the option -oldcxx, then everything links, but then TPad->Divide() > doesn't work. > > So, my belief is that the CXX 6.2 compiler is adding references to the > functions: > __vec_new_eh > __vec_delete > when it compiles which cannot be resolved when trying to link with the f90 > linker. > > So, how do I make it stop doing that? Or, where are these functions in the > system libraries so that I can include them? > > Paul > > >Hi Paul, > >The CXX and CXX6.2 versions have been compiled with the same options > >on two different machines. I did not check if the internal compiler > >default > >options were different however. > > > >I had several requests to provide a separate distribution for CXX6.2 > >(apparently there are differences in the way exception handling is > >supported > >in the two versions). > > > >I am a bit disapointed by the evolution of the cxx compiler on alpha. > >I see more and more people moving to egcs on thsi platform. > > > >Rene > > Paul Nord wrote: > > > > Hi Rene, > > > > Yes, isn't it interesting! Our collaboration has made the big step toward > > modern programming techniques with F90. > > > > I find no rrti option for F90, unfortunatly (nor rtti, for that matter). > > However, I do note that I can compile just fine with the CXX6 release of ROOT > > rather than the CXX6.2 release. What, may I ask, is the difference? And, will > > you continue to support both compiler versions? And if the CXX6 version will > > compile with CXX6.2, why are there two version of ROOT? > > > > Of course, the simple solution will probably be to link with CXX.... but that > > may be more of a political matter. > > > > Thanks for the prompt and accurate reply, > > Paul > > > > Rene Brun wrote: > > >Hi Paul, > > >Interesting! You are the first user that I know calling Root from > > >a F90 program. > > > > > >I assume that you are running on alpha/osf and use the cxx compiler. > > >Root version 2.23 is compiled with the option -nostdnew > > >The rrti option is activated by default on this compiler. > > > > > >I have no idea of the equivalent options with f90. > > > > > >Rene Brun > > > > Paul Nord wrote: > > > > > > Good Day, > > > > > > Got this link error. Any ideas? > > > > > > f90 -o mofia20 -O -Olimit 1500 -extend_source > > > -L/usr/users/pnord/e614soft/triumf/cfm/2.2/main > > > -L/usr/users/pnord/e614soft/triumf/bru/1.1 > > > -L/usr/users/pnord/e614soft/triumf/ybos/4.1/library > > > -L/usr/users/pnord/e614soft/triumf/mofia/2.0/lib > > > -L/usr/users/pnord/e614soft/gplot -L/gluona/cernlib/98/lib > > > -L/usr/users/e614/root/lib > > > /usr/users/pnord/e614soft/triumf/mofia/2.0/source/main/mofia.o precision_mod.o > > > det_geom_mod.o calibrations_mod.o chambers_mod.o tdcmap_mod.o tdc_mod.o > > > filters_mod.o hists_mod.o user_hists_mod.o dplot.o func.o hist.o -lmain > > > -lphoto -L/usr/users/e614/root/lib -lNew -lCore -lCint -lHist -lGraf -lGraf3d > > > -lGpad -lTree -lRint -lPostscript -lMatrix -lPhysics -lm > > > -L/usr/users/e614/root/lib -lNew -lCore -lCint -lHist -lGraf -lGraf3d -lGpad > > > -lTree -lRint -lPostscript -lMatrix -lPhysics -lGui -lm -lcxx -lX3d -lttf -lcfm > > > -lybos -lbru -lgplot -lpacklib -lmathlib -lkernlib -ldummy > > > ld: > > > Unresolved: > > > __vtbl_3std9type_info__3std > > > __vec_new_eh > > > __vec_delete > > > f90: Severe: Failed while trying to link. > > > gmake: *** [mofia] Error 1 > > > > > > Paul Nord
This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:23 MET