heller.umass-cs%udel-relay@sri-unix.UUCP (06/01/83)
From: Robert LISPer DMHeller <heller.umass-cs@udel-relay> About the p-System FORTRAN77 on the Sage II: I have used it and have found the following: 1) it only supports 16-bit integers. (ie Integer*2) 2) the boolean operators .AND., .OR., & .NOT. only work on LOGICAL expressions, unlike all other versions of FORTRAN77 I have encountered (DEC VAX/VMS, CDC Cyber NOS, even DEC XVM (PDP-15)!). 3) there is a "byte sex" bug in the intrinsic function CHAR (bytes are swaped: ie ix.eq.ichar(char(ix*256)) ) 4) it does not support sub-string expressions 5) it won't alow reading or writing non-CHARACTER typed variables using Aw format specs. 6) it does not support Xw or Ow format specs. 7) it does not (as far as i can tell) support hex or octal constants. 8) it does not alow assigning a character constant to a non-CHARACTER variable. I have not tried assigning .TRUE. or .FALSE. to an integer variable, but I would not be suprised if the compiler complains. 9) it does not allow equivalenving CHARACTER vars with non-CHARACTER vars 10) its support of separate compilation is definately non-standard - that is modules which compile fine under other FORTRAN compilers (see above) may not compile under the p-System FORTRAN compiler without modifications. I have not read the "...Fortran 77 (actually 78) ANSI STD manual, as the official Fortran Subset.", so I don't know if the above is in fact the standard. I have also heard (from Sage) that neither the p-System FORTRAN 77 nor the p-System BASIC will link with Pascal code. They (Sage) are not even sure the FORTRAN or BASIC will link with assembly language. The Sage people also warned me not to write any large programs in FORTRAN 77 because it is buggy and is a poor implementation. It seems that SoftTech just sort of threw the FORTRAN 77 & BASIC in as an afterthought. That is, the p-System is good for Pascal and slight dithering in assembly language. Also, the p-System version of Pascal is a toy version (compared to what is available on mainframes or even for CP/M-80 & CP/M-86 (ie Pascal-MT+, etc.). It does not run as fast as you would like (oh it is alot faster than p-code on an Apple, but the Sage II does have a fast 16-bit processor in it...) I have switch to CP/M-68K ($250.00 U.S.A from Sage). The C is very close to UNIX C (but does not support float types). 32-bit integers are available. C programs written under UNIX should compile & run under CP/M-68K C. While the system as it comes from DR & Sage does not have a screen editor, MINCE (being written in C) has been ported to CP/M-68K, so all is not lost... I have it on good advice that UNIX is not worth bothering with unless you have a LARGE disk (at least 10-20 Meg.) -- there is lots of stuff in UNIX - it just won't fit on a 5 Meg disk. Oh, you can get a striped down version - that is missing many (usefull!) utilities, but you might as well have CP/M-68K. (I know of a friend who has a WICAT w/ a 5 Meg hard disk and had UNIX up, but did not have room for all of the header files! - CP/M-68K comes with a full set and fits on three 5.25" DSDD 80-track floppies!) robert heller