ted@imsvax.UUCP (04/19/86)
Steve Schlaifer of the Jet Propulsion Labs writes: >In the reference, the author complains (bitterly :-)) about the microsoft >MSDOS Fortran compiler. After looking at what was available, we chose the >Lahey Fortran compiler for our use in porting a set of large libraries from >our mainframes to micros. The main reason we chose Lahey was that it >supported the full F77 ANSI standard not just the subset. We have since >found fewer bugs in the compiler than the people around here who chose the >Microsoft compiler and what we feel is a much more professional attitude toward >the user. Take a look and I think you will like it. Half of what you see in mainframe Fortran manuals are things which go OVER the 77 standard, and these break two or three ways, including: 1. Things such as character variables being stored in integer or real arrays needed to make your 35 year base of software compatible. Without this, a Fortran compiler is of little use, and this isn't part of the 77 standard. 2. Things such as namelist reads which one or two of the mainframe manufacturers couldn't deal with and isn't in the standard, but are none- theless in general use in the mainframe community and would be needed for many reasonable applications involving porting mainframe software to micros. 3. All of your normal debugging stuff etc.; this isn't part of the standard either. Most micro Fortrans amount to a literal rendition of the 77 standard or less, which obviously doesn't get it. The serious Fortrans for Micros are the Absoft for 68000 equipment and the Lahey for 808x based equipment. The Lahey compiler is rumored to have been funded in its development by one of the companies marketing large statistical packages which deemed the MicroSoft and Ryan-McFarland compilers inadequate for the PC version of their product.