 |
 |
Hello, It's true that f77 tended to have more extensions than f90/95 has. Perhaps this is because f90/95 is a "bigger" language, and compiler vendors have had their plates full making highly optimized f95 compilers. Personally, I've come to appreciate the greater degree of standardization one gets with f95. This certainly reduces the issue of platform-specific extensions. However, what one loses is the "laboratory of many vendors" whence many of the f90/95 features came. Let me attempt to split a hair: I've always appreciated VAX Fortran as being one of the best compilers around. However, the _langauge_ which VAX Fortran supported is one of the sloppiest around. Can everyone remember, without checking, whether its even valued integers which are .true.? Or the other way 'round? Vendor extensions aren't a Fortran specific issue. How many proprietary C libraries or pragmas or outright extensions are there? (Rspecially given C's desire to be "close to the HW"?) I agree interlanguage communication is a valuable area for standardization. John P Grimes wrote: > <snip> > I do however have 1 huge complaint. This has to do with the large > number of DEC extensions to F77 that are platform specific. Programmers > used them in desperation in F77 and then kept using them unecessarily on > DEC machines with F90. This caused highly platform specific code. I > don't know who to blame but its ugly. Also, in C you have long int (any > platform), in Fortran its integer(kind=8) or integer(kind=4) on 32 bit > machines. Good planning gets around this problem but thats not what we > had. This is especially a problem when pasing parameters to C. <snip again> -- Cheers! Dan Nagle [log in to unmask] %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%