NEWS.POS Some features already there: * various startup codes for baserelative and resident code * compared to the p.OS MainCode.c, you have a correct argv[0] here. * auto-open library feature (even before commercial compilers ?) * some ANSI function are already working :) * nice bugs (who knows in a beta version ?) Test this in real condition as much as you can. (Remember to take care, crashes are always possible in a beta) I haven't many time until June, so any help is welcome. Now for the progress report: * All startups should be ok now. After many many crashes I understand that it was critical to save a5, so I do that now :) (really a f*cking bug!) (very probably p.OS use a5 as a frame pointer) You should not encounter crashes caused by libnix-pos (but take care!) * Library and device startup code are not ready yet. Anyway you should play with the examples given with p.OS. * Standard library: Strange outputs with printf, when using fp numbers (dont forget to link with -lm if you want to try this). Depending on the format precision (%.2f seems ok), garbage characters will appear in the floating part: like "2.3-/:5" Where do these strange characters come from ? This bug occurs only when not using -m68881. It seems to be independant from the types (float/double), and from the output format (f,g). May come from nix/stdio/vfprintf.c, nix/math/*.c, math/trans/*.c Use -m68881 if you want correct output. not available for p.OS: (as there was some work there ;) nix/locale/setlocale.c nix/time/strftime.c nix/extra/dirent.c nix/extra/fstat.c nix_main/__nocommandline.c (higher priorities than old "main(char *)" ) misc/detach.c misc/swapstack.c stubs/__cpucheck.c still use AmigaOS ExecBase (anyone has a clue for p.OS?) libstack.a is a dummy library yet, don't use (If 10 persons fix one source each, all is done :) I think that's enough... Probably there is much more things to fix, because I'm sure that exciting new bugs have been introduced. 27-Apr-97 lecorfec@univ-mlv.fr