1 LibF77 News - most recent first 2 3August 9, 1985: 4 add d_dprod() for dprod() when -r8 flag is used. 5 6July 6, 1985: 7 switch r_lg10 and d_lg10 to use new log10 in -lm. 8 switch functions of form r_*.c to be declared as float instead of double. 9 use -f flag for most single precision routines to get around C bug; 10 also put extra stores in 4 routines to bypass bug (See Usef in 11 Makefile & file Test_float). 12 13June 26, 1985: 14 move f77_abort() from libF77/main.c to libI77/f77_abort.c 15 so C progs can call f77 subroutines. 16 fix pow_ii to divide by zero for i**j where i=0 and j is < 0 17 fix pow_hh by applying all fixes from pow_ii to it. 18 19June 6, 1985, 4.3BSD beta version: 20 all error exits now go through f77_abort(). Normally get dump 21 only if loaded with -lg (via f77 -g). Can use f77_dump_flag 22 envir. var. to override that (set it to 'y' or 'n'). 23 24 new routine: random() 25 26 delete sinh.o, tanh.o - identical to C routines in libm 27 cabs.o, d_sqrt.o, z_abs.o - now in assembler in libm 28 bit.o: not documented, man 3f bit now shows better ways. 29 routines in bit.o also had inverse arg. order to built in funct. 30 31 complex division by zero - now get error comment. 32