SUN/IRAF V2.3 Installation, 13 June 1986. Sun-2, software floating point -------------------------------------------------------------------------------- /usr/iraf/ Ran rmbin on bin, sys, pkg, math, lib, local, unix. Unpacked source-only archive from master VAX/UNIX system "lyra" of 12-June-86 17:00. (6/13) unix/as/ unix/as/ishift.s unix/as/zsvjmp.s Deleted VAX contents of as/, copied mc68000/ishift.SUN and zsvjmp.SUN to as/*.s. (6/13) unix/hlib/[dir]1mach.f Commented out pdp (vax) data statements and uncommented IEEE statements. (6/13) unix/hlib/mach.h unix/hlib/libc/spp.h Replaced EPSILONR and EPSILOND with values from unix/mc68000/README. Set both BYTE_SWAPs to NO in mach.h. (6/13) unix/hlib/mkiraf.csh Edited iraf root to new pathname. (6/13) unix/hlib/mkpkg.inc Set USE_CALCOMP to "no". (6/13) unix/hlib/libc/iraf.h Changed iraf root pathname. (6/13) unix/hlib/libc/kernel.h Commented out DEBUGMEM (i.e. use standard UNIX malloc()) in order that spp programs can call SunCGI. (6/13) unix/ Bootstrap (sh -x mkpkg.csh); no problems. (6/13) unix/hlib/alloc.e As su, "/etc/chown 0 alloc.e" so it can allocate devices. (6/13) /usr/bin/alloc -> $hlib/alloc.e /usr/bin/generic -> $hlib/generic.e /usr/bin/mkpkg -> $hlib/mkpkg.e /usr/bin/rmbin -> $hlib/rmbin.e /usr/bin/rmfiles -> $hlib/rmfiles.e /usr/bin/rtar -> $hlib/rtar.e /usr/bin/wtar -> $hlib/wtar.e /usr/bin/xc -> $hlib/xc.e /usr/bin/cl -> $hlib/cl.e Made sure these symbolic links existed. (6/13) -------------------------------------------------------------------------------- Bootstrap is complete. Hand-compile the Sun f77 optimizer-bug routines before performing the full recompile sysgen. sys/fio/fstati.x F77 runtime math/bevington/matinv.f F77 runtime sys/gio/nspp/portlib/gridal.f F77 runtime sys/ki/irafks.x F77 runtime noao/twodspec/longslit/transform/igsfit/igsfit.f F77 runtime All these files have to be hand-compiled without optimization from previous experience with the Sun V2.0 f77 compiler; none of these files nor their .x parents have changed since last built on a Sun-2. (6/13) iraf/ Began full compile & link sysgen at 16:33. (6/13) fio/fioclean.x Fixed null pointer bug. (6/14) bin/x_display.e bin/x_cv.e images/tv/cv/libpkg.a images/tv/display/libpkg.a Removed to free up disk space. (6/14) $iraf Unpacked an incremental archive from lyra! to bring us current to 09:15 June 20, 1986. Performed an incremental sysgen from the root. (6/20) noao/twodspec/apextract/* The preceding sysgen failed to find all files in icfit/* except for 3 source files. To be safe, I (sr) made a complete archive of apextract from lyra at 10:20, unpacked it, and did a "mkpkg update" from here. (6/20) pkg/images/geometry/t_imtranspose.x F77 runtime f77 optimizer bug was causing "Pixel subscript out of bounds" in imtranspose. Compiled without optimization from cl with "cl> !mkpkg xflags=-c; cd ..; mkpkg update". (6/20) noao/twodspec/apexold/* Because of old file dates and because this directory on lyra was created only by renaming the directory itself, it didn't get copied in the incremental tar earlier this week. Made a fresh archive on lyra and a "mkpkg update". (6/20) $iraf Unpacked a lyra! incremental archive taken at 14:40 June 21; performed a root incremental sysgen. (6/21) $iraf Unpacked a lyra! incremental archive taken at 08:10 June 23; performed a root incremental sysgen. (6/23) noao/lib/scr/* noao/twodspec/apexold/* noao/twodspec/apextract/* Replaced with full archives from lyra! to attempt to clear up bugs possibly related to renaming, rather than copying, files on lyra! since the June 12 full archive -- these could be missed by the incremental archives (find . -mtime -N). MKPKG updates in both apextract directories. (6/23) pkg/cl/unop.c Changed call to sscanf (sval, "%f", &rresult) in case OP_REAL to sscanf (sval, "%lf", &rresult) because on SUNs, only the first four bytes starting at rresult were being written into (rresult is declared "double"). Should be changed on master system as well. (6/23) -------------------------------------------------------------------------------- June 24, 1986: Extensive testing for several days has revealed no further bugs particular to the SUN-2 in its implementation at NOAO (SUN V2.0, no fpa, no array processor). Although we do not have a 1/2" tape drive on the SUN-2, the SUN-3's have a known bug which appears to be SUN's problem -- I/O error when trying to backspace-record when the drive is positioned between the two EOF marks written at close time. It is probable we will have a workaround for this bug in early July. --------------------------------------------------------------------------------