-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Dirk Eddelbuettel wrote: > Could be the compiler, could be R. FWIW the debian/rules specifies gcc 4.2 > as per > > > # edd 03 Apr 2006 switch to gfortran > # edd 04 Apr 2006 use optimflags, build per-compiler flags later > # edd 26 Jun 2007 need 4.2 for fortran linking > fortrancompiler = F77=gfortran-4.2 > compiler = gcc-4.2 > cxxcompiler = g++-4.2 > optimflags = -O3 -pipe > > > and those variables are used later in the configure and make calls. > > Another possible source of trouble could be if you mix local and external > packages with different compilers, though it's been a while that I've seen > trouble from that. But maybe life is trickier on amd64 -- dunno.
Okay. I have only a few packages that are from CRAN (as opposed to the Debian repos), but the only packages that are loaded at startup are the basic ones that load out of the box on R. Comparing the output to starting R through valgrind on 32bit Debian, the problem seems to be early in the startup sequence. On the 32bit, there is lots of valgrind output when base, methods, etc. are loaded that the 64bit evidently never gets to in my case. Thanks, Ben -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQFGylRNzQDSXIcN85kRAj0EAJ9IGMtkuv+u/qkXtYN21U49nhjx1ACfU7Bv K7fk5jjKjXuckT6Pbw+IE1E= =9oyz -----END PGP SIGNATURE----- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
