On Sun, Jun 13, 2010 at 11:32:33PM +0200, Steven Mestdagh wrote: > Landry Breuil [2010-06-13, 13:08:56]: > > On Sun, Jun 13, 2010 at 10:24:09AM +0200, Steven Mestdagh wrote: > > > hi, > > > > > > this is a standalone port of gfortran, to be used on archs which have gcc4 > > > in base. this port is based on the gcc/4.2 port. > > > > > > gfortran compiles both fortran 77 and fortran 95 code. > > > > > > tested ok on amd64. if someone can try it on sparc64, i'd appreciate it. > > > comments on the port welcome of course. > > > > Seems the fpic libs don't build on sparc64. I can send you the buildlog > > if needed.. > > > > Error in package: > > /usr/ports/pobj/gfortran-4.2.4/fake-sparc64/usr/local/lib/fpic/libgfortran.so.2.0 > > does not exist > > Let's try without that; I've put in the PIC fragment like in the gcc port.
Builds and packages fine. I've built blas/cblas with the fortran.port.mk diff, and it correctly installed gfortran port. Make regress fails though : ===> Regression check for cblas-1.0 gfortran -o xscblat1 c_sblat1.o c_sblas1.o -L/usr/ports/pobj/cblas-1.0/CBLAS/src -lcblas -L/usr/local/lib -lblas /usr/local/lib/libgfortran.so.2.0: warning: strcpy() is almost always misused, please use strlcpy() /usr/local/lib/libgfortran.so.2.0: warning: strcat() is almost always misused, please use strlcat() /usr/local/lib/libgfortran.so.2.0: warning: sprintf() is often misused, please use snprintf() c_sblat1.o(.text+0x20): In function `MAIN__': : undefined reference to `s_wsfe' c_sblat1.o(.text+0x28): In function `MAIN__': : undefined reference to `e_wsfe' c_sblat1.o(.text+0xe8): In function `MAIN__': : undefined reference to `s_wsfe' c_sblat1.o(.text+0xf0): In function `MAIN__': : undefined reference to `e_wsfe' c_sblat1.o(.text+0x168): In function `MAIN__': : undefined reference to `s_stop' c_sblat1.o(.text+0x190): In function `header_': : undefined reference to `s_wsfe' c_sblat1.o(.text+0x1b8): In function `header_': : undefined reference to `do_fio' c_sblat1.o(.text+0x1fc): In function `header_': ..... c_sblat1.o(.text+0x17d0): more undefined references to `do_fio' follow c_sblat1.o(.text+0x17f8): In function `itest1_': : undefined reference to `e_wsfe' c_sblat1.o(.text+0x1808): In function `itest1_': : undefined reference to `s_wsfe' c_sblat1.o(.text+0x1810): In function `itest1_': : undefined reference to `e_wsfe' c_sblat1.o(.text+0x1824): In function `itest1_': : undefined reference to `s_wsfe' c_sblat1.o(.text+0x182c): In function `itest1_': : undefined reference to `e_wsfe' /usr/local/lib/libblas.so.1.0: undefined reference to `z_abs' /usr/local/lib/libblas.so.1.0: undefined reference to `c_abs' *** Error code 1 Stop in /usr/ports/pobj/cblas-1.0/CBLAS/testing (line 73 of Makefile). etc etc etc A bunch of errors during py-numpy configure too, dunno if new: C compiler: cc -DNDEBUG -O2 -pipe -DTHREAD_STACK_SIZE=0x20000 -fPIC -O2 -pipe -Wall -fPIC compile options: '-Inumpy/core/src -Inumpy/core/include -I/usr/local/include/python2.5 -c' cc: _configtest.c _configtest.c:1: warning: conflicting types for built-in function 'modfl' cc _configtest.o -lm -o _configtest _configtest.o(.text+0x4): In function `main': : undefined reference to `modfl' collect2: ld returned 1 exit status _configtest.o(.text+0x4): In function `main': : undefined reference to `modfl' collect2: ld returned 1 exit status failure. there's no regress test for blas itself. For py-numpy regress: Ran 2030 tests in 53.287s FAILED (KNOWNFAIL=2, SKIP=12, failures=30) Landry