sandrock@aries.scs.uiuc.edu (Mark Sandrock) (04/23/91)
Has anyone succeeded in building the NCAR Graphics 3.00 distribution under the RISC/os? I have been trying on an M120 with RISC/os 4.51 and the NCAR 3.00 distribution. If I use the default (SYSV) build, it produces an undefined symbol: bcopy in cgmtrans. If I do a BSD43 build, everything compiles, but I get fatal runtime errors such that it appears that various arguments are not being passed correctly. Any help would be appreciated! Mark Sandrock -- BITNET: sandrock@uiucscs Univ. of Illinois at Urbana-Champaign Internet: sandrock@aries.scs.uiuc.edu Chemical Sciences Computing Services Voice: 217-244-0561 505 S. Mathews Ave., Urbana, IL 61801
sandrock@aries.scs.uiuc.edu (Mark Sandrock) (04/27/91)
sandrock@aries.scs.uiuc.edu (Mark Sandrock) writes: >Has anyone succeeded in building the NCAR Graphics 3.00 distribution >under the RISC/os? I have been trying on an M120 with RISC/os 4.51 >and the NCAR 3.00 distribution. If I use the default (SYSV) build, >it produces an undefined symbol: bcopy in cgmtrans. If I do a BSD43 >build, everything compiles, but I get fatal runtime errors such that >it appears that various arguments are not being passed correctly. Someone suggested added -lbsd to resolve the undefined bcopy message, which did work, but the runtime errors under both sysv and bsd43 were caused by a different problem. The NCAR GKS library makes use of many Fortran COMMON statements, and also contains several BLOCKDATA files which initialize many COMMON variables. To force the loading of the BLOCKDATA modules from the GKS library, the GKS open routine contains an EXTERNAL statement invoking the names of the two BLOCKDATA modules. This method presumbably works under other os's, but does not appear to work under RISC/os 4.51 (with the current f77/ld, etc.). I was able to get the BLOCKDATA modules to be loaded by added CALL statements to the two BLOCKDATA modules, but this is not a very neat solution to me... Does anyone know of a cleaner way around this problem? Is it something basic that I'm overlooking perhaps? Thanks in advance. Mark Sandrock -- BITNET: sandrock@uiucscs Univ. of Illinois at Urbana-Champaign Internet: sandrock@aries.scs.uiuc.edu Chemical Sciences Computing Services Voice: 217-244-0561 505 S. Mathews Ave., Urbana, IL 61801