lepreau@utah-cs.UUCP (Jay Lepreau) (07/17/84)
I dun forgot: as 4.2 was distributed csh used its own alloc.c, which is the old one, and of course does exhibit that nasty behavior after awhile. I had changed it here to use a modification of the standard libc one. Get rid of stdio, add support for the "alloc" builtin (use the mstats routine) and one surprise: remove the ASSERT for freeing a free block at the beginning of free(). Turns out csh does exactly that when processing the =~ and !~ operators, and if you look real close you'll discover that check is commented out in the original alloc.c. Jay Lepreau