ndeng@EULER.BERKELEY.EDU (05/15/91)
Subject: Summary:"Wanted: Information on 386 Compilers for Code distribution." Newsgroups: comp.lang.fortran In article <9105032158.AA18558@euler.Berkeley.EDU> I write: > >Hi, fellow Fortran netters. I am looking for the information about currently >available 386 FORTRAN compilers, their advantage/disadvantages and reviews, >and especially, the requirement(s) to build .EXE files for distribution. >(Do I need some special license for this, like NDP, or it is bundled in the >compiler package, like Lahey? etc. etc.) I would appreciate if anybody can >shear his/her personal/business experience on this regard. > >Please respond by email to reduce bandwidth. If this topic generates enough >interest, I'll post a summary. > >Mike >ndeng@euler.berkeley.edu I received 5 responses to my inquiry. Two messages referred to the recent article writen by Ross Taylor, which was originally posted in this group, then refined and published in the recent issue of the FORTRAN Journal. One message recommended the Lahey F77L-EM/32, and another reported problems about WATCOM 386. The last one suggested an alternative of using MS-FORTRAN with Ergo OS/286 DOS extender, which can have upto 16 MB of memory. In addition, I have digged out all the old emails and made phone calls to various companies to find out all the details. The following is a brief summary: There are six different FORTRAN 386 compilers I am aware of currently. They are NDP-FORTRAN 386 WATCOM 386 SVS-F77 386 FTN FORTRAN77-386 LPI FORTRAN 77 Laney F77L-EM/32 NDP FORTRAN uses their own NDP-run drivers. If one wants to distribute codes , one needs to buy a OS/386 from Ergo ($695) plus a runtime license ($500 for the first executable, $400 for the second, and $300 each if you have five or more executables for a period of three years). The NDP compiler costs $895. It is also compatible with the PharLap DOS Extender which cost $495 for the development version, and a 1000 copy runtime license costs $1995 (infinite time). The Phar Lap runtime license is for infinite time but needs one to submit quarterly auditting report to the company. Once you sell over 1,000 copies, you need to pay for 2% of the list price of your code or $10 (Whichever is greater) royalty to the company. WATCOM 386 and SVS-F77 386 cost $895 and $795 respectively. They both use the PharLap DOS Extender so one has to spend the $495 plus $1995 = $2490 in order to distribute the code. FTN FORTRAN77-386 sells for somewhat $1100 and charges about $100 for each runtime kernel to be distributed with the code. It uses itself's DOS extender. However, according to Ross Taylor, the company expect the end user to buy their own copy of dos extender, and doesn't like to grant the runtime license. I didn't get any information about LPI77. Finally, the Lahey F77L-EM/32. It sells for $995 list and uses a reduced version of OS/386 ($395). Or one can also use the full version of Ergo OS/386 for $695 (Of course!). It comes with unlimited runtime license. One thing particularly good about Lahey is that, though I've been read this newsgroup and various computer magazines for over three years, I haven't seen any serious criticism about Lahey. Whereas all other compilers were somewhat criticized. (Correct me if I am wrong and I'll be grateful before I throw my money away!) One thing NOT good about Lahey is that, the company seems to try to milk people in a step-by-step way. If one wants to add some DOS functions or do some graphics in the code, one needs to buy separate libraries, but most of these added features are provided in other, say NDP, compilers. Also, WATCOM, NDP and SVS all have their own C compilers, so mixed language programming is not a problem. But Lahey has only an interface with Metaware High C. Nevertheless, Lahey appears to be the choice. Besides its excellent rating about the compiling time and documentations, it cost, if one include the expenses of the runtime license, seems to be less than most of the others. And one can always find a software clearinghouse to get some discount. Many thanks to all people who provided the insights about various compilers and all the information in this message. Mike ndeng@euler.berkeley.edu