ras@sgfb.ssd.ray.com (Ralph A. Shaw) (06/28/90)
Last month, we finally took the plunge and upgraded from OSx 4.1 (yep, that old) to OSx 5.0C on our 9820. Since then, several annoying things were discovered, some of which we've called in as trouble-reports, others we just copied over the old binary from 4.1. I'm posting this to see if others had run into them as well... (As preface, our users are mainly using UCB universe, with the AT&T Toolchest's ksh88 Korn Shell installed as /bin/sh) 1) Fortran has this peculiar habit of producing code that doesn't seem to *DO* anything. Most of the non-trivial existing fortran programs that our users have brought to my attention compile without errors, and run without results. Interactive programs don't interact, and other programs don't produce any output. In one case a 1200 line program did manage to get /usr/lib/fcomp to core dump, but that was all. This was called into Pyramid, and a tape was sent a few weeks ago. In the mean time I've reverted back to the OSx 4.1 compiler, and things work OK... (I was told that there weren't any known problems with fortran - does anyone actually use it?) 2) The "script" program seems to have a problem with CR/LF and acts like its in cbreak or raw mode or something. This might be an interaction with ksh88, but we didn't have this problem under 4.1... Fixed by going back to old binary. 3) The "calendar" shell script, which basically pipes the passwd file into a "while read arg" loop, was mangling shell variables. This was definitely something screwy with ksh88, as the normal /bin/sh works fine. 4) ditroff does not behave sanely for either the -man or -ms macros, producing output that consists of a single line, and a message from ditroff about "line too long". (This is the ditroff borrowed from the "att" universe). [As a side note, has anyone noticed that the "man" pages source (troff input) is no longer included on the system? It seems that its turned into a "layered product"... Probably not worth paying for, but an annoyance just the same]. Some of these problems may be self-induced (non-standard shell and all), but they are a pain. As I type, a PTF tape is reported to be on the way... What bearing that has on these problems remains to be seen. Ralph Shaw Raytheon Co., Submarine Signal Division ras@sgfb.ssd.ray.com -- Ralph UUCP-Mail raybed2!ras ...and from Raytheon Corporate Standard CDC-Mail GRA2+RAS,CDFQ+RAS