BattenIG%CS.BHAM.AC.UK@MITVMA.MIT.EDU (Ian G Batten) (10/05/88)
I'm just mounting emacs (18.51) on the machines at my new place of employment. These are straightforward 68020 machines with Unisoft V.2 (swapping) soon to be 68030 with V.3 (paging). I'm using an m-file that is essentially m-dual.h with s-unipl5-2.h. The compliation proceeds without any problems, apart from: (1) FIONREAD is in sys/ioctl.h so that has to be included by keyboard.c. (2) It seems to have the same interface to utime as "IRIS" ifdefs in sysdep.c, so an extra "|| defined(FULCRUM)" is needed there. temacs links correctly, and if invoked directly runs fine. The screen clears and shows a mode line, esc-x displays "M-x" correctly and typing "?" splits the screen and shows the completions available. Typing into what is presumably "*scratch*" works fine too. However, attempts at "./temacs -batch -l inc-vers" cause a core-dump at line 198 in lread.c. This is the first strncpy in openp. sdb (gdb port is next week) interestingly doesn't show the stack frame associated with openp, havin g the strncpy immediately descended from the Fload. Fload seems to be getting sensible arguments, but I'm about to insert debugging printfs and recompile. I first suspected the alloca of the "fn" target of the strncpy, but replacing the call to alloca with malloc made no difference. Does anyone have any helpful suggestions? Any Unisoft users? Help! Thanks in advance. ian