Path: senator-bedfellow.mit.edu!enterpoop.mit.edu!usc!cs.utexas.edu!uunet!olivea!sun-barr!news2me.EBay.Sun.COM!exodus.Eng.Sun.COM!appserv.Eng.Sun.COM!appserv!khb From: khb@chiba.Eng.Sun.COM (Keith Bierman-khb@chiba.eng.sun.com::SunPro) Newsgroups: comp.lang.fortran,comp.answers,news.answers Subject: Fortran FAQ Followup-To: comp.lang.fortran Date: 25 May 93 16:16:12 Organization: SunPro Lines: 1461 Approved: news-answers-request@MIT.Edu Distribution: world Message-ID: NNTP-Posting-Host: chiba Xref: senator-bedfellow.mit.edu comp.lang.fortran:13993 comp.answers:800 news.answers:8788 Archive-name: fortran-faq Frequency: mostly monthly Here are some answers to frequently asked questions. The "author", as is the custom, has appropriated posted responses as seemed apt. I have tried to leave attributions in, as correctly as possible. To anyone who has been offended by omission or otherwise, my apologies. I shall give priority to corrections regarding attribution. No one takes responsibility for any of this text, neither the employer of the "author", the "author", friends of the "author", pets of the "author" nor anyone else. Your mileage WILL vary. If you have comments/suggestions/edit proposals please send them to me (keith.bierman@eng.sun.com). I do not promise to accept 'em. I encourage others to make better faq lists, so I can retire this one. ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; summary of changes: q2) updated q14) location of cfortran.h specified. q28) added ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Questions: Q0) How should one spell FORTRAN/Fortran? Q1) Where can I get a copy of the Fortran 90 Standard? How about electronic copies? Q2) What Fortran 90 translators are available? Q3) I have heard of fortran "lints" what are they, and where can I get one? Q4) "Why do people use FORTRAN?, C is so much better" Q5) Why can't I get an electronic version of the standard? Q6) Where can I get "foo" (some random package) Q7) Where can I get a free compiler? Q8) What is the best compiler for a PC? Q9) How does Fortran 90 relate to FORTRAN '77 and what is Fortran 90? Q10) My compiler is mis-behaving; who enforces the standard? Q11) What are good books on Fortran? Q12) Are there pretty printers for FORTRAN? Q13) Why are there aimless debates? Q14) How do I call f77 from C (and visa versa) Q15) What constitutes good FORTRAN style? Q16) For whatever reasons, I want to translate my Fortran into C. What tools are available? Q17) For whatever reasons, I want to translate my existing C code into Fortran. What tools are available? Q18) What is preprocessing, how can it help? How can it hurt? Q19) How can I convert an existing FORTRAN 77 program to the free form source of Fortran 90? Q20) Who creates these silly standards anyway? Q21) How can I read my VAX binary data somewhere else? Q22) My F77 program compiled ok on a , but gives me heaps of syntax errors on a . What's wrong? Q23) My F77 program ran ok on a , but on a it just gives me strange results. What's wrong? Q24) Is there a WEB for Fortran (and what is web anyway)? Q25) Where can I find coded BLAS (and what are coded BLAS?) Q26) How common is DO ... END DO? Q27) Where can I learn more about the history of Fortran? Q28) Fortran text editors? ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;; Q0) How should one spell FORTRAN/Fortran? FORTRAN is generally the preferred spelling for discussions of versions of the language prior to the current one ("90"). Fortran is the spelling chosen by X3J3 and WG5. In this document a feeble effort has been made to capitalize accordingly (e.g. vast existing software ... FORTRAN vs. generic Fortran to mean all versions of the standard, and specifically the modern dialect, ISO 1539:1991). From: walt@netcom.com (Walt Brainerd) There was an effort to "standardize" on spelling of programming languages just after F77 became a standard. The rule: if you say the letters, it is all caps (APL); if you pronounce it as a word, it is not (Cobol, Fortran, Ada). See, for example the difinitive article describing Fortran 77 in the Oct 1978 issue of the Comm. of the ACM. The timing was such that FORTRAN got put on the standard itself, though many always after that have referred to it as Fortran 77. Of course, there are those who think it is not truly Fortran if not written with all caps. ISO 1539:1991 and its ANSI counterpart X3.198-1992 consistently employ the spelling "Fortran" to refer to the language being defined. Reference(s) to the older version employ "small caps" for the "ORTRAN" characters. Q1) Where can I get a copy of the Fortran 90 Standard? You can't get it from ftp. The various standards bodies get a substantial fraction of their income from sales of standards documents. This is unfortunate Electronic versions are available, see the note from Walt below. From: julian@cernvax.cern.ch (julian bunn) The new Fortran 90 standard has now been formally published and is available from the ISO central secretariat or from the national member bodies. The price is about $120. Some ordering detail follow. I can supply addresses in other countries not mentioned on request (to METCALF@CERNVM). The document reference is ISO/IEC 1539 : 1991. There is no ISBN. ISO 1, rue de Varembe Case postale 56 1211 Geneva 20 Switzerland Fax: +41 22 734 10 79 ANSI 1430 Broadway New York, NY 10018 BSI 2 Park Street London W1A 2BS DIN Burggrafenstrasse 6 Postfach 1107 D-1000 Berlin 30 AFNOR Tour Europe Cedex 7 92049 Paris La Defence SCC 1200-45 O'Connor Ottawa Ontario K1P 6N7 Mike Metcalf (metcalf@cernvm) and from walt@netcom.com For those of you who don't read comp.lang.fortran, you might be interested to know that I (walt brainard) just called Global Engineering to see if the sell ISO 1539:1991. They do; the price is $225. And You can obtain copies through: Global Engineering Documents 2805 McGaw Ave. Irvine, CA. 92714 (714) 261-1455 (800) 854-7179 New ANSI number ANSI X3.198-1992 How about electronic copies? Return-Path: walt@netcom.com Subject: ISO 1539 in electronic form In accordance with an official agreement with the International Standards Organization, Unicomp is now able to distribute electronic versions of the Fortran 90 standard: "ISO/IEC 1539 : 1991, Information technology--Programming languages--Fortran". The money received from this effort will go partly to fund ISO activities and partly to recover the costs incurred by Unicomp in preparing and typesetting the standard document. The prices are set by ISO. The document can be obtained in three versions: 1. An ascii version suitable for viewing on a computer terminal using any kind of editor. Cost: USD 125. 2. A PostScript version with a license allowing the purchaser to print n paper copies. Cost: USD 125 + 10n. 3. Complete source in ditroff with macros and software to extract and create the annexes. The source constitutes a fairly high level marked up document; for example, each program beginning and ending is marked and there are few low-level typographic commands such as size and font changes. Cost USD 1000. I am quite enthused especially about version (2). If you want to have 10 copies for your organization, and it costs $10 to make a printed copy, then the cost to make the 10 copies would be $125 + $200, or just $32.50 per copy, which is a substantial savings over purchasing paper copies. Versions (1) and (3) will be accompanied by a license restricting use to one CPU and prohibiting copying, except for backup purposes, etc. The version (2) license will prohibit distributing any of the printed copies outside of the purchasing organization. If you have special requirements, such as wanting to distribute a copy with each version of your compiler or using the source as a part of your documentation, we can make special arrangements, subject to the approval of the ISO. Please advise me of your requirements and we can work up a proposal together. ISO and Unicomp think this will provide the often requested access to the standard in electronic form. This is the first time this is being tried, so we hope that organizations will be careful to observe the rules and encourage the continued availability of this and other standards in electronic form. Payment can be made by Visa or MasterCard, or with a check on a US Bank in US funds. We will accept a purchase order only if the amount is $500 or more. Walter S. Brainerd; Unicomp; phone: (415) 949-4052 email: walt@netcom.com; fax: (415) 949-4058; Q2) What Fortran 90 translators are available? From: schumanm@Informatik.TU-Muenchen.DE (Matthias Schumann) From: macphed@dvinci.USask.Ca (Ian MacPhedran) This is the promised summary of my request for information about Fortran 90 (f90) compilers. I am sorry for the delay in getting this posted, but I was away for a week, and still haven't caught up. I received a large amount of information, including a previous summary. Rather than parroting the entire list, I have prepared the following breakdown of the points which interested me. ------------------------------------------------------------------------------ Product: NAGware Fortran 90 Compiler Descript: Compiles f90 code to C source, or directly to executable file under MS-DOS. Therefore requires native C compiler as back end on most platforms. Platform: Apollo DN10000, DECstations, HP 9000, IBM RS/6000, NeXT, Intel 386/486 (MSDOS), Sun 3, Sun 4 (SunOS 4.1) VAX/VMS forthcoming. Pricing: Varies with platform. Academic Pricing available. Site License available +------------------------------------------------------------+ | NAGWare F90 Compiler | +------------------------------------------------------------+ | License Fee Options for Initial Computer | +------------------------------------------------------------+ | | Standard | Academic | +--------+-------------------------+-------------------------+ | | Perpetual| Maintenance| Perpetual| Maintenance| | | Fee | Fee | Fee | Fee | +--------+-----------+-------------+-----------+-------------+ | CLASS 1| $895 | $160 | $537 | $97 | (MSDOS) +--------+-----------+-------------+-------------------------+ | CLASS 2| $1,495 | $270 | $897 | $162 | (Low end work- +--------+-----------+-------------+-----------+-------------+ stations) | CLASS 3| $1,995 | $360 | $1,197 | $215 | +--------+-----------+-------------+-----------+-------------+ | CLASS 4| $2,495 | $450 | $1,497 | $270 | (Server class +--------+-----------+-------------+-----------+-------------+ computers) | CLASS 5| $3,295 | $594 | $1,977 | $356 | +--------+-----------+-------------+-----------+-------------+ | CLASS 6| $4,295 | $774 | $2,577 | $464 | +--------+-----------+-------------+-----------+-------------+ | CLASS 7| $5,295 | $945 | $3,177 | $572 | +--------+-----------+-------------+-----------+-------------+ (US Dollars) Contact: Sheila Caswell, Account Manager, Numerical Algorithms Group, Inc. 1400 Opus Place, Suite 200 Downers Grove, IL, USA 60515-5702 Phone: (708) 971-2337 Source: E-mail messages, postings, and paper mailing ------------------------------------------------------------------------------ Product: VAST-90 Descript: Converts f77 code to f90, and reverse. Will "de-spaghettify" f77 code to a more structured form. Requires native f77 compiler to compile f90 programs. Platform: SPARC (Sun 4), IBM RS/6000 Pricing: Single Machine: starts at $1900, Sun, IBM HP and DEC workstations. Higher prices for larger machines. Included: VAST-90, f90 driver, f90 library and documentation. Volume discounts and site license available. Contact: Pacific-Sierra Research Corp. Switchboard: (310) 314-2300 2901 28th Street Fax: (310) 314-2323 Santa Monica CA 90405 VAST-90 e-mail: dave@psrv.com ------------------------------------------------------------------------------ Product: PF90 Version 2.0 Descript: Converts f90 programs to f77, then compiles with native f77 compiler. Therefore native f77 compiler is required. Platform: Sun 4, IBM RS/6000, Silicon Graphics, DECstations, Convex, Alliant, IBM 3090 (AIX), Cray Pricing: Sun, RS/6000, SGI, DECstation: $2000 Convex, Alliant: $5000; IBM 3090, Cray: $10000 (US Dollars, single machine, multiuser license, must purchase f77 compiler separately.) Contact: ParaSoft Corporation 2500 E. Foothill Blvd Pasadena, CA 91107 Phone: (818) 792-9941 E-mail: f90-info@parasoft.com Source: E-mail messages, and PostScript advertisement ------------------------------------------------------------------------------ Product: PV version 1.1 of Fortran 90 compiler Descript: Port of NAG's f90 compiler to MS-DOS, together with Salford Software Ltd. Platform: Intel 386/486 Computers Pricing: $1950.00 (single user, US currency) Contact: OTG Systems Inc Phone: (717) 222-9100 Source: E-mail message/summary of previous queries ------------------------------------------------------------------------------ The following is from a previous summary from David Bernholdt: Convex is reported to have an incomplete f90 compiler in beta. Other compilers in preparation will be available: EPC (Edinburgh Portable Compilers) and Lahey late 1992 Cray Research early 1993 DEC for VMS late 1993 ------------------------------------------------------------------------------ I wish to thank the following people for their help, and apologize to any which I have missed. psrv!bob@uu.psi.com naginc!caswell@sunbird.Central.Sun.COM malcolm@nag.co.uk roman@parasoft.com khb@chiba.Eng.Sun.COM METCALF@crnvma.cern.ch ---------------------------------------------------------------------------- Ian MacPhedran, Engineering Computer Centre, University of Saskatchewan. 2B13 Engineering Building, U. of S. Campus, Saskatoon, Sask., CANADA S7N 0W0 macphed@dvinci.USask.CA macphedran@sask.USask.CA Walt Brainerd walt@netcom.com: Yes, a Fortran 90 system can be had for $134. Salford FTN90 is a full PC Fortran 90 implementation based on the front end developed by the Numerical Algorithms Group, Ltd. The Entry Level system requires that all parts of a program reside in one file (except those that are brought in with an INCLUDE line). That file is compiled and executed without producing a permanent executable file. It runs on a 386SX, 386DX, or 486 system with DOS 3.3 or later. Contact Unicomp for ordering information in North America. +1-415-949-4052 4058 (fax) To find out the price and how to get it outside of North America, contact Salford Software, Venables Building, 5 Cockcroft Road, Salford, M5 4NT, United Kingdom, +44 61 745 5678, +44 61 745 5666 (fax). >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> Q3) I have heard of fortran "lints" what are they, and where can I get one? Fortran compilers are not required (and most do not) to protect users from themselves, that is to say call sub(1) ... subroutine sub(i) i=i+10 return end Is non-standard complying but the compiler need not tell you about it. Arguably worse, the compiler can produce any result it deems fit, up to and including the start of world war III (assuming the right optional hardware has been installed). Fortran lint programs focus on searching out search programming errors and alerting the user. There are a wide variety of commercial and pd products. A partial list follows: IPT FORTRAN lint 1096 East Meadow Circle, Palo Alto, CA 94303, 415/494-7500. Forwarn from quibus 719 527 1384. QA-FORTRAN from PRL UK 01 942 9242 and perhaps (817) 589 0949 FORCHECK from Leiden University ND 31 71 276804 FTNCHEK from netlib. One may receive FTNCHEK by mailing to netlib@ornl.gov the following message: send ftnchek from fortran And Ajay Shah, (213)749-8133, ajayshah@usc.edu You may want to mention f2c followed by an ANSI C compiler as a great free fortran lint. I could not have survived writing fortran if it were not for f2c. You need to know some C to deal with the error messages, but that's not a bad price to pay. From: "John D. McCalpin" Not too many people use it, but you might want to add TOOLPACK to the list of FORTRAN "lint" packages. Here is the blurb that I send to people who are interested: ====================================================================== TOOLPACK is a large set of utilities written in FORTRAN to do FORTRAN code analysis and transformation. TOOLPACK begins by actually parsing the program with a fully FORTRAN- compliant parser and then does all the code transformations on the parse and comment trees. This means that you cannot confuse TOOLPACK by silly FORTRAN-isms that can easily overwhelm more naive utilities (such as ftnchek). For example, TOOLPACK correctly handles statements such as: D O U B L E P R E C I S I O N A(100,100) doubleprecision fred DO I = 1.2 READ(I,J) = 10. The usual interface to TOOLPACK is a set of 'csh' scripts that run the various TOOLPACK utilities to do specific tasks. The ones you will find most useful are: pol, polx, apt, dapt, decs, getlst, and discard. These are the prettyprinter, precision converter, declaration standardizer, and some necessary utilities. See below for more details. By aware that TOOLPACK is *very* rigorous about the FORTRAN-77 standard. You may or may not consider this an advantage. The scripts are described by the 'scripts' script, which delivers the following: ---------------------------------------------------------------------------- In the Toolpack script summaries below, the section numbers refer to the Unix environment Users' Guide where more detailed information can be found. To obtain on-line information about a script's usage, type its name without any arguments. FORTRAN Analysis Facilities getlst Produce a listing showing statement and token numbers. Report lexical scanning warnings and errors. (3.1.1) syn Report errors and warnings detected by lexical scanning, parsing, and examining a set of symbol attributes. (3.1.2) sem Report errors and warnings detected by lexical scanning, parsing, and examining an extended set of symbol attributes. (3.1.3) pfort Report errors and warnings detected by lexical scanning, parsing, examining an extended set of symbol attributes, checking for unsafe references, and checking conformance to a portable subset of FORTRAN. (3.1.4) statdoc In a user-supplied report template, place information derived from static analysis to assist in documenting the program. Examples of information the user may request are COMMON block usage, symbol attributes, and a graph of subprogram calls. (3.1.5) inst Instrument a FORTRAN program so that the instrumented program, when executed, produces information about program execution. (3.2.1) rundoc Execute a program instrumented by inst and, in a user-supplied report template, place information derived from dynamic analysis to assist in documenting the program. For example, the user may determine the frequency of execution of program segments. (3.2.2) FORTRAN Transformation Facilities pol Format a FORTRAN program under control of user-supplied options in a Polish option file. (4.1.1) polx Construct a Polish option file via a menu-driven editor. (4.1.2) decs Rebuild the declarative part of a FORTRAN program. (4.2.1) apt Transform a single-precision version of a FORTRAN program to double precision or vice versa. (4.3.1) dapt Convert precision and rebuild the declarations, combining the functions of apt and decs. (4.3.2) cname Change the names in a FORTRAN program that satisfy conditions derived from information in either the lexical token stream or the symbol table or both. (4.4.1) lname Transform a FORTRAN program containing long names to a program with standard names. (4.4.2) stf Rebuild the flow of control in a FORTRAN program to standardized form. (4.5.1) ucs Transform nests of DO loops matching certain paradigms so that the transformed code executes more efficiently on vector machines. (4.6.1) Miscellaneous Facilities fdiff Compare two FORTRAN programs at the lexical token level. (5.1.1) dac Compare two data files, neglecting certain formatting differences and numerical differences smaller than a given tolerance. (5.1.2) vcon Create, edit, and retrieve versions of a file contained in a version file. (5.2.1) discard Remove unneeded files created by the above scripts. (5.3.1) scripts Print this summary of the scripts. (5.4.1) ---------------------------------------------------------------------------- TOOLPACK can be obtained by anonymous ftp from perelandra.cms.udel.edu in the directory pub/Lang/Toolpack/. The UNIX version is in the file toolpack.tar.Z Serious hackers will want the versions in the Distrib directory. The package is known to compile on Silicon Graphics and Sun machines, though the Makefile is reasonably braindamaged. You will need about 16 MB for the stripped executables on an SGI machine. Figure on double that for compiling the package..... Have fun! -- John D. McCalpin mccalpin@perelandra.cms.udel.edu Assistant Professor mccalpin@brahms.udel.edu College of Marine Studies, U. Del. John.McCalpin@mvs.udel.edu Q4) "Why do people use FORTRAN?, C is so much better" One should avoid such pointless arguments .... if you feel this way, don't subscribe to comp.lang.fortran. This sort of question always triggers off a protracted discussion, the highlights of which are typically: a) FORTRAN and C have different semantics. A FORTRAN optimizer knows more about aliasing, function interactions, and I/O. A C optimizer has to infer or compute such information. C bigots typically have neither written such optimizers nor worked with folks who do it for a living, and are prone to dismiss such arguments as being petty and neolithic. FORTRAN programmers are often a bit more in touch with high performance computing, and are unwilling to bet that heavily on compiler wizardry. In addition, ANSI/ISO C has very tight requirements on expression evaluation (an overreaction to the K&R ignore parenthesis days) which cripples an optimizer even further. The restrictions are spelled out in Section 5.1.2.3 of ISO/IEC 9899-1990. Both the '77 and '90 Fortran standards allow the optimizer more latitude. Of course, compiler options, vendor extensions and the like sometimes narrow the difference. Also, some Fortran compilers do not take advantage of all of these features of the language, thus narrowing the gap in the other direction. On any given code, on any given system, all bets are off. b) There is a vast body of existing FORTRAN code (much of which is publically available and of high quality). Numerical codes are particularly difficult to "vet", scientific establishments usually do not have large otherwise idle programming staffs, etc. so massive recoding into any new language is typically resisted quite strongly. c) Fortran tends to meet some of the needs of scientists better. Most notably, it has built in support for: - variable dimension array arguments in subroutines - a compiler-supported infix exponentiation operator which is generic with respect to both precision and type, *and* which is generally handled very efficiently or the commonly occuring special case floating-point**small-integer - complex arithmetic - generic-precision intrinsic functions d) Retraining staff is quite expensive. e) It is sometimes argued that: - Jonathan Thornburg Fortran tends to be easier for non-experts to learn than C, because its "mental model of the computer" is much simpler. For example, in Fortran the programmer can generally avoid learning about pointers and memory addresses, while these are essential in C. More generally, in Fortran the difference between (C notation) x, &x, and often even *x is basically hidden, while in C it's exposed. For non-experts this makes Fortran a considerably simpler language. Because of this relative simplicity, for simple programming tasks which fall within its domain, (say writing a simple least-squares fitting routine), Fortran generally requires much less computer science knowledge of the programmer than C does, and is thus much easier to use. Q5) Why can't I get an electronic version of the standard? Someday, perhaps you can. However, at the moment the various standards bodies finance some fraction of their work via sales of standards documents. ISO and ANSI have both claimed copyright to the finished standard, so those who would make and distribute copies should be forewarned. It should be noted that various people are working to correct this; but it is unlikely to ever become free. Q6) Where can I get "foo" (some random package) Q6) Where can I get "foo" (some random package) contributed by: Marc R. Roussel mroussel@alchemy.chem.utoronto.ca Use archie. archie is an online database of what is available and where on the "net". Archie can be used either via telnet or by mail. For information about archie send mail archie@archie.rutgers.edu subject: help And you will get back copious directions on how to use archie. If you prefer interactive experimentation, telnet to archie.rutgers.edu and log in as archie. No password will be required. The first thing you will be shown is a list of other archie servers. If one of these servers is geographically much closer to you than Rutgers, NJ, please disconnect (by typing 'quit') and use that one. To save you this step, here is the latest list of archie sites: archie.rutgers.edu 128.6.18.15 (Rutgers University) archie.unl.edu 129.93.1.14 (University of Nebraska in Lincoln) archie.sura.net 128.167.254.179 (SURAnet archie server) archie.ans.net 147.225.1.2 (ANS archie server) archie.au 139.130.4.6 (Australian server) archie.funet.fi 128.214.6.100 (European server in Finland) archie.doc.ic.ac.uk 146.169.11.3 (UK/England server) archie.cs.huji.ac.il 132.65.6.15 (Israel server) archie.wide.ad.jp 133.4.3.6 (Japanese server) I believe that all of these servers run the mail server as well as the telnet and archie server software. (The archie server is a third way to use archie. It's probably the best way, but it requires that you install software.) Once you logged into an archie server, you will want to make sure that all the information you will retrieve will be mailed to you. Type set mailto userid@machine.foo.edu where, of course, you will substitute your own email address for the made-up one shown above. You are now ready to search the database. If at any time you want to know what options are available to you, type help. To search for a program or file, type prog foo where foo is the name of the program or file required. Once your search is done, type 'mail' to have the output sent to you. (The output will almost always be several pages long.) Then type 'quit' to exit. Q7) Where can I get a free compiler? One such is f2c. From: mwm@a.gp.cs.cmu.edu (Mark Maimone) Newsgroups: comp.lang.fortran,comp.lang.c Subject: Re: Fortran to C translator Keywords: fortran, C, translator, f2c Date: 13 Sep 90 01:43:30 GMT Organization: Carnegie-Mellon University, CS/RI Since there have been several requests for a Fortran to C translator in the past week, I'm reposting the announcement about f2c. The short answer is you can get f2c by anonymous-ftp from research.att.com in directory dist/f2c. ------------------------------------------------------------------------------ Source for f2c, a Fortran 77 to C translator jointly developed by folks from Bell Labs, Bellcore, and Carnegie Mellon, is now freely available. F2c was derived from the original UNIX operating system's f77(1), and the generated C follows f77's calling conventions; on some machines, the resulting object files are interchangeable with (and behave indistinguishably from) objects compiled by f77. The main "advantage" of f2c is that it converts ANSI standard Fortran 77 into C without manual intervention, at least when invoked by a suitable script or makefile (that may need to exercise an f2c option to ensure that COMMON blocks are defined just once). The main "problems" are that f2c does no code restructuring (e.g., gotos are preserved) and that Fortran I/O gets converted into a bunch of calls; thus the translated C code doesn't look too pretty, and in general one would need to maintain the Fortran rather than its translation into C. [F2c is not meant to displace the services of commercial vendors whose business is to convert Fortran into maintainable C.] There is a plethora of options, many of which exist to support different compilation environments for the translated C (e.g., ANSI C or C++ compatibility, different type sizes, separate files for COMMON blocks to appease "smart" linkers). So far f2c (and f2c-generated source) has compiled successfully on many machines: Sun, Vax, IBMRT, Apollo, SGI, MIPS, and Cray to name a few. F2c has been under test by the net community for over a year and has been verified on the NBS tests, several large math libraries, floating point tests, even code for laying cable on the ocean floor! To find out about f2c, send the following E-mail message to netlib (netlib@research.att.com or research!netlib): send index from f2c Your message will be answered automatically (by a program -- see CACM vol. 30 #5 (May, 1987), pp. 403-407). You will receive a reply explaining how to automatically acquire f2c source (about 600K), f2c library source (130K), and supporting info (man page, etc). Or you can anonymous-FTP to research.att.com and look in directory dist/f2c at these files: all.Z -- 250K compressed shar file for f2c f2c.ps.Z -- 24 page tech report describing f2c index -- general info about files libf77.Z, libi77.Z -- compressed shar files for libraries ****************************** DISCLAIMER ****************************** Careful! Anything free comes with no guarantee. ************************************************************************ ---------------------------------------------------------------------- Mark Maimone phone: (412) 268 - 7698 Carnegie Mellon Computer Science email: mwm@cs.cmu.edu cmcspt!mwm@cs.cmu.edu Notes: f2c accepts only fairly vanilla FORTRAN; vendor supplied f77's usually produce better quality code, and accept a wider variety of codes. Q8) What is the best compiler for a PC? There are many products, some are quite good. Few are free. f2c and gcc can be had for the PC environment. Popular ones are: Lahey (very fast compilation; excellent reputation for support) Watcom (touted for good optimization) MicroWay (support for odd floating point units) LPI (multi-platform support) MicroSoft (various good hooks into windows and such) Absoft (multi-platform support) It is beyond the scope of a faq to provide a commercial endorsement. Q9) How does Fortran 90 relate to FORTRAN '77? With a few minor exceptions, Fortran 90 is a superset of X3.9-1978 FORTRAN. But this does not mean that all "77" codes will port sans changes. Many (if not most) programmers employed constructs beyond the '77 standard, or rely on unspecified behavior (say, assuming that an OPEN of an existing file will position the file pointer to just past the last record already written) which has changed (that is to say, has become specified). This leads to the obvious question, what is new in Fortran 90? A complete answer would require considerable text. Some of the most obvious additions are: 1) array notation (operators, etc.) 2) dynamic memory allocation 3) derived types and operator overloading 4) keyword argument passing, INTENT (in, out, inout) 5) modules 6) modern control structures 7) free format source code form 8) other stuff While it is always tricky to characterize the motives of a large group of people, I am inclined to try as follows: '90 incorporates two sets of improvements: (1) relatively minor fixups that *could* have been done earlier (2) relatively major changes to enable better software engineering practices. Sometimes a "minor" fixup has major effect, such as addition of free form source form combined with canonization of the MIL-STD 1753 INCLUDE. I further go off on a limb and assert that it was the goal of the *committee* to evolve Fortran in a fashion to enable it to continue to be the premier language for scientific computation. Q10) My compiler is mis-behaving; who enforces the standard? ANSI and ISO standards do not usually have a particular enforcement mechanism. Local bodies sometimes do. However, it should be borne in mind that if *your* source code is not standard compliant there is *NO* obligation for a FORTRAN ('77 and before) compiler to do *ANYTHING* in particular. In Fortran (90) the text in "constraints" must be tested and a warning produced (compiler option can be used to evade this, of course). Some (notably Guy Steele, with respect to another standard) have noted that when non-standard complying code is encountered, a compiler may do *ANYTHING* including initation of Global Warfare. Keep this in mind. When you do find a bona fide compiler bug, you are generally best served by reporting it to the *vendor*. If you neglect to tell the vendor, how can you complain about it not being fixed? When reporting a *suspected* bug be sure to be quite specific about the computer system, operating system rev level (patches applied if known) and *compiler*version* (and patches thereof). It is very hard for people to read your mind; but they will try. The attempts are often entertaining, sometimes helpful, but always an inefficient use of people-time and net-bandwidth. Also note that it is generally helpful if you cut down the example to the smallest size you can. Vendors are developers too; the tendency is invest time/money where one can get the biggest bang for the buck. Q11) What are good books on Fortran? Don't know if they are good. Inclusion in the list is not endorsement. Have misplaced name of the compiler of the original list Author Title Year Kruger Efficient Fortran Programming 1990 Mojena/Ageloff FORTRAN 77 1990 Metcalf/Reid FORTRAN 90 EXPLAINED 1990 Boyle FORTRAN 77 PDQ 1989 Bezner FORTRAN 77 1989 Tremblay PROGRAMMING IN FORTRAN 77 1988 Salmon ... ENGINEERS & SCIENTISTS WITH FORTRAN 77 1988 Nyhoff/Leestma FORTRAN 77 FOR ENGINEERS & SCIENTISTS 1988 McCracken/Salmon ... ENGINEERS & SCIENTISTS WITH FORTRAN 77 1988 Davis/Hoffman FORTRAN 77: A STRUCTURED DISCIPLINED STYLE 1988 Barnard/Skillicorn FORTRAN 77 FOR ENGINEERS AND SCIENTISTS 1988 Mashaw PROGRAMMING STRUCTURED FORTRAN 77 1987 Cole FORTRAN 77: A STRUCTURED ... APPROACH 1987 Boillot UNDERSTANDING FORTRAN-77 1987 Starkey/Ross FUNDAMENTAL PROGRAMMING WITH FORTRAN 77 1986 Rouse/Bugnitz INTRODUCTION TO FORTRAN 77 1986 Ratzer FORTRAN 77 COURSE 1986 Page FORTRAN 77 FOR HUMANS 1986 Lehman ... SOCIAL SCIENCES: ALGORITHMS & FORTRAN 77 1986 Smith FORTRAN 77: A PROBLEM-SOLVING APPROACH 1985 Shelly FORTRAN 77: AN INTRODUCTION 1985 Nickerson FUNDAMENTALS OF FORTRAN 77 PROGRAMMING 1985 Metcalf EFFECTIVE FORTRAN 77 1985 McKeown STRUCTURED PROGRAMMING USING FORTRAN 77 1985 Hume FORTRAN 77 FOR SCIENTISTS & ENGINEERS 1985 Dillman PROBLEM SOLVING WITH FORTRAN 77 1985 Brainerd FORTRAN 77 FUNDAMENTALS AND STYLE 1985 Borse FORTRAN 77 & NUMERICAL METHODS FOR ENGINEERS1985 Adman FORTRAN 77 SOLUTIONS NON-SCIENTIFIC PROBS. 1985 SSI, Inc. Staff FORTRAN 77 REFERENCE 1984 Etter PROBLEM SOLVING WITH STRUCTURED FORTRAN 77 1984 Etter PROBLEM SOLVING USING FORTRAN 77 ? Dyck FORTRAN 77: A STRUCTURED APPROACH ... 1984 Chivers/Clark FORTRAN 77: A HANDS ON APPROACH 1984 Adman FORTRAN 77 FOR NON-SCIENTISTS 1984 Willamson/Levesque * A GUIDEBOOK TO FORTRAN ON SUPERCOMPUTER 1989 Rule FORTRAN 77: A PRACTICAL APPROACH 1983 Rouse/Bugnitz PROGRAMMING THE IBM PC: FORTRAN 77 1983 Nyhoff/Leestma PROBLEM SOLVING WITH FORTRAN 77 1983 Marateck FORTRAN 77 1983 Lehmnkuhl FORTRAN 77 1983 Law ANSI FORTRAN 77: INTRO. TO SOFTWARE DESIGN 1983 Holoien/Behforooz ... STRUCTURED PROGRAMMING WITH FORTRAN 77 1983 Grout FUNDAMENTAL ... PROGRAMMING USING FORTRAN 77 1983 Fleming/Redish THE U. S. MC MASTER GLOSSARY OF FORTRAN-77 1983 Cole ANSI FORTRAN IV WITH FORTRAN 77 EXTENSIONS 1983 Wu ANSI FORTRAN IV & 77 AND BUSINESS PROGRAMS 1982 Pollack STRUCTURED FORTRAN 77 PROGRAMMING 1982 Katzan FORTRAN 77 1982 Gibson/Young INTRODUCTION TO PROGRAMMING USING FORTRAN 77 1982 Ellis STRUCTURED APPROACH FORTRAN 77 PROGRAMMING 1982 Durgin FORTRAN 77 1982 Nanney A PROBLEM-SOLVING APPROACH USING FORTRAN 77 1981 Merchant FORTRAN 77: LANGUAGE AND STYLE 1981 Khailany BUSINESS PROGRAMMING FORTRAN IV/ANSI FORTRAN 1981 Ashcroft PROGRAMMING WITH FORTRAN 77 1981 Wagener FORTRAN 77 ? Wagener PRINCIPLES OF FORTRAN 77 PROGRAMMING 1980 Meissner/Organick FORTRAN 77 FEATURING STRUCTURED PROGRAMMING 1980 Hume/Holt PROGRAMMING FORTRAN 77 1979 Balfour PROGRAMMING IN STANDARD FORTRAN 77 1979 Brainerd Programmer's guide to Fortran 90 1990 Adams et al Fortran 90 Handbook 1991 Counihan Fortran 90 1991 O'Reilly&Assoc Unix for FORTRAN Programmers 1990 Q12) Are there pretty printers for FORTRAN? Yes. One such is SPAG: OTG voice 717 222 9100 fax 717 229 9103 Authors are Polyhedron Software Ltd Magdalen House 98 Abingdon Road Standlake Witney Oxon OX8 7RN Tel 0865 300 579 part of their plusFORT product. Another is Fortran development Tools from Quibus 714 527 1384 Also FOR_STRUC from cobalt-blue. 404 518 1116 These more than pretty print, they optionally restructure your code (duplicating code as needed to tidy up strange GOTO lists, turning them into IF-THEN chains, and etc.). From: dappel@grafted.UUCP (Dave Appel) Newsgroups: comp.lang.fortran Subject: Re: code beautifier wanted Date: 4 Mar 92 21:30:25 GMT Organization: GRAFTED, Central Indiana's Usenet BBS 317-881-4369 wg@cbnewsm.att.com (Bill Gieske) writes: > I have tons of old FORTRAN code, most of it in upper case, the majority > of it not indented. Is there a code beautifier, either PD or $$ that I > can run the code through to improve the readability, hence the main- > tainability? > Reply to me direct. I will summarize if appropriate. > Bill Gieske > AT&T Bell Laboratories > wg%alux2@att.research.com Call "The Connection at 800-336-1166" and ask for their software catalog. In their catalog that I have, Sprint 1992, there is an ad on page 39 from AutoCASE Technology. They have a product called "AutoFLOW-FORTRAN" that lists for $1,995. It claims to automatically document your existing source code. AutoCASE's number is 408-446-2273. On page 93 is an ad from POWERLINE Software Inc. Their number is 800-257-5773, 206-623-9204. They offer a product called SOURCE PRINT+, which they call a code management tool with "multi-style formatting with structured code blocking." They have a Fortran/Basic version for $169, and a Delux Multi-language version for $249. This program is for DOS (MS-DOS presumably) Windows, and OS/2. Hope this helps. -- Dave Appel The Grafted Branch BBS 317-881-4369 internet: dappel@grafted.UUCP uucp: ..!uunet!grafted.UUCP!dappel -= newsfeeds available, contact robert@towers.rn.com =- Hindsight/Fortran Summary: Features include the ability to draw an interactive structure chart and display coverage, software complexity and performance information on the structure chart. There are also code tracing features, for instance 3 logic diagrams are produced. Common blocks can be traced through the structure charts, including specific variables within common blocks. Hindsight is excellent for documentation, code inspection, and bringing new software engineers up to speed on new code. For more information or a free demo copy, contact: Dan Zimmerman, National Account Manager Advanced Software Automation, Inc. 3130A Coronado Dr Santa Clara, 95054 Phone: 408 492-1668 Fax: 408 492-1669 Email: zimme@hindsight.com Alternative contact (if Dan cannot be reached: manny@hindsight.com (John Mansperger) Hindsight/FORTRAN Version 1.0 features Hindsight/FORTRAN Version 1.0 is a fully integrated software maintenance, testing, and re-engineering environment which addresses many problems of the software development cycle, from design to maintenance. Hindsight/FORTRAN Version 1.0 parses FORTRAN source code, decomposes system structure, and from this analysis, produces a single database from which all of Hindsight's charts, diagrams and reports are produced. The results of Hindsight's analysis is a detailed picture of the static structure of the software system, a representation of the common and local data, and connections to the dynamically collected run-time data. From this, Hindsight generates active structure charts where the user can isolate logical subtrees to analyze both the parents and the children of a selected function. Hindsight collects and analyzes performance, complexity, and test coverage data. With the structure chart, the user can display this data in a graphical and easy to use format. From the structure chart, the user can edit multiple functions at a time without having to specify the file or directory path where the file exists. Hindsight even displays the use of common blocks on the structure chart. Lastly, Hindsight generates logic diagrams for the entire program, provides point-and-click tracing between high-level and detailed logic with active line numbers, and generates over fifty detailed reports. All capabilities mentioned above are integrated in a menu driven environment utilizing the OpenLook or Motif Graphical User Interface. Engineers only need to specify a list of source files to be analyzed and select the operation and parameters for their purposes. Hindsight provides the ability for multiple diagrams and reports to be produced simultaneously during a single Hindsight session. Hindsight/FORTRAN Version 1.0 provides an innovative technique to graphically represent a software system. The structure chart reflects the system structure, from which the user can get a clear overview of the program being examined. Hindsight also allows the user to view or edit the original code with a click on any subprogram box in the structure chart. A major time-saving advance is Hindsight's ability to reanalyze an edited function and update the structural database without reanalyzing the entire system. The J-Diagram and ActionPlus diagram are two graphical representations of the logic in a software system and can contain a single module or the entire system. Both diagrams utilize the active line number technique to link high-level module calls with detailed module logic. Hindsight also actively links computed GOTO statements with their destination labels by clicking on the label. Users can walk through programs and trace the logic without having to memorize the directories and files where subprograms are defined and called. This technique saves a great deal of the users time. Hindsight/FORTRAN Version 1.0 has many integrated automated tools to generate documents either on-screen or on paper. Because the documents are automatically generated, they can be updated as soon as engineers have modified their source code. This approach guarantees the consistency between the design documents and the source code. The updated and correct documents will be of great value in maintaining and improving software products. These documents are available on-line to the user and readily accessible through Hindsight's interactive report mechanism. Hindsight Version 1.0 has the ability to provide segment or line test coverage data. Hindsight generates test coverage data and displays the data on the structure chart in a statistical way. Hindsight can also display the precise location of testing deficiencies on the J-Diagram, which helps engineers find the exact part of the program where the code has not been tested. Hindsight's ability to display performance data, complexity metrics, and test coverage data on charts and diagrams play an important role in controlling program quality. Hindsight/FORTRAN Version 1.0 uses a new notation for the J-diagram which accommodates branch test coverage capability. The new notation has a better and more precise method of showing control levels, complexity, and segment number counting. The branch test coverage data generation and display is a great enhancement for test coverage analysis. Hindsight's branch test coverage is able to detect invisible branches, finding more problems that would otherwise be hidden in programs. With this information, users can add test cases for covering the untested code. Hindsight/FORTRAN Version 1.0 contains the ability to speed up the development process by generating a specific order for coding the modules of a users system. By analyzing a group of modules containing nothing more than empty shells of subprograms that call the other subprograms in the system, Hindsight generates the coding order for the modules to ensure that the process will be testable with the use of a single driver process and without the need to stub out uncoded subprograms. The bottom-up coding approach saves time that would be wasted coding unneeded stubs. As development proceeds, Hindsight's ability to verify that calls to other subprograms are utilizing the proper interface will also save time and improve program quality. Other tools: Refine/FORTRAN ("re-engineering tool") Reasoning Systems Inc 415 494 6201 FORM (graphical interface to toolpack) University of Kent 44 227 762811 pub/fortran/fxref.tar.Z from biome.bio.ns.ca. create cross reference of a FORTRAN program. CLEAN77 available from Purdue University. Mail Greg Flint (afc.klaatu.cc.purdue.edu) for more info. There is a program called TIDY that is available via FTP from simtel20 pub/msdos/fortran/tidy621.zip on oak.oakland.edu Q13) Why are there aimless debates? There are some persistent individuals who would like computer systems to work in a fashion somewhat unlike they do now. It seems pointless to debate with them; the debates (in various groups) always take the same form and never result in meaningful dialog. One can usually recognize such by reading a week or so's worth of postings. Sometimes you may have to read for a month to recognize such. When you do, please try to avoid triggering (or responding to) such individuals. Many (if not most) newsreaders support a feature entitled KILL files. As you recognize individuals, you may wish to put them into your local kill file. For example, the style employed by GNUS: (gnus-kill "From" "hrubin@pop.stat.purdue.edu") (gnus-kill "Subject" "test") the first kills off all postings from the named individual, the second kills off all postings including the beloved "test" string. See your local NewsGod for details suitable for your local system. Q14) How do I call f77 from C (and visa versa) This is quite platform dependent. For Suns see the FORTRAN User's Guide, Chapter 14. There is a package available from usenet which attempts to make this "quick and easy" for a wide range of platforms: ftp.uu.net (137.39.1.9) Location: /comp.sources.misc/volume20 DIRECTORY rwxr-xr-x 512 Jun 27 1991 cfortran It is on many other sites (around the world) too. See archie if you need other pointers. For some systems, you have to initialize a runtime system explicitly if you call a different language, and stopping execution from the other program may not work. The most recent version of cfortran.h is available via anon. ftp at zebra.desy.de. Q15) What constitutes good FORTRAN style? One rendition of a FORTRAN 77 style guide is available through anonymous ftp from ics.uci.edu (128.195.1.1). To retrieve (please note that it's not really "anonymous", that's just the Name that you'll be using): % ftp ics.uci.edu anonymous cd pub/levine ascii get F77_Style_Guide bye If you can't access this site directly, please send an e-mail request to levine@ics.uci.edu (BITNET: levine@uci, UUCP: ...!uunet!ucivax!levine). Q16) For whatever reasons, I want to translate my Fortran into C. What tools are available? f90 from NAG, see above f2c see above. FORTRAN=C=FORTRIX=Rapitech Rapitech (914) 368-3000 FORTRAN=C=FOR_C=Cobalt Blue 404 518 1116 FORTRAN=C=PROMULA.FORTRAN=Promula (614) 263-5512 Q17) For whatever reasons, I want to translate my existing C code into Fortran. What tools are available? Regretably none. This is indeed unfortunate, as even a limited translator could help with typical C header files. Q18) What is preprocessing, how can it help? How can it hurt? Preprocessing often refers to usage of a Macro-prepressor upon ones source code prior to compilation. How can this help? It can make it easier to move code between machines. How can this hurt? It can cause difficulties in compilation of the processed code; most (if not all) macro-preprocessors know nothing about Fortran syntax and code layout rules. So, a common problem is that after preprocessing some text may be lost past "sacred" column 72. Those working with compilers that optionally compile *past* line 72 should probably employ that option when using a preprocessor. Common prepressors are: cpp, m4, and ratfor. Many sites have their own custom prepressors. Basic functions are: 1) definition of a symbol 2) conditional code selection based on a symbol 3) substition of a symbol by its definition Traditional BSD Unix f77 processors treat a file named foo.F (as opposed to foo.f) as one that should be run through cpp prior to compilation "automatically". Q19) How can I convert an existing FORTRAN 77 program to the free form source of Fortran 90? A program to do this, CONVERT, may be obtained by sending a request to metcalf@cernvm.cern.ch. Q20) Who creates these silly standards anyway? Typically X3J3. X3J3 is an ANSI subcommittee dedicated to Fortran. WG5 is the ISO counterpart. WG5 owns responsibility for Fortran on an international basis. WG5 has previously tasked X3J3 to do the work. This arrangement continues. WG5 is composed of Fortran users, vendors, and academics from several ISO supporting nations. Delegates represent *their*countries* not their companies; so several delegates from a single company is permitted. ANSI rules prohibit multiple voting delegates from the same company. X3J3 is composed of users (aerospace, government labs, military, DECUS, railroads, oil to name a few), vendors (IBM, CRI, Sun, Convex, DEC, UNISYS, to name a few) and the odd academic (oxford, yale, liverpool, to name a couple). Members need not be US citizens nor must their company be US domiciled. Being a member of a standards group is typically involves non-trivial work. To be effective, one should plan on at least 8 weeks of time per year (those who are really doing the hard work do far more). This time commitment is typically far more expensive than the travel and membership costs. X3J3 meetings are open to the public. There are typically 4 meetings a year, typically 3 are in the US and 1 *may* be overseas (to precede or follow the WG5 plenary session). Membership fees are levied by ANSI, and are on the near order of $400 (plus an extra $300ish for those who wish to be members of the US delegation to ISO). In addition, attendees to a particular X3J3 meeting pay a "meeting fee" which covers reproduction costs, snacks and etc. WG5 has established various goals and targets for future work. Roughly speaking 5yrs rather than 13years are the targets for future work. Current work projects include cleanup and interpretations of Fortran (90), features for future versions of the standard (e.g. parallel processing, "object-oriented" technologies, etc.). In addition to work done directly by X3J3, there is work on standardized modules, and OS bindings taking place in other organizations. X3J3 would like to keep track of such efforts, those involved are invited to inform X3J3 early in their development efforts if possible. X3J3 is currently working with X3H5, DIN (varying string character) and tracking the efforts of HPFF. New members are always welcome. Visitors are also; though it is very hard to get a good grip on things in only one meeting! Contact the X3J3 chair or vice-chair for more information: jwagener@trc.amoco.com chair mbsh@holstein.fc.hp.com vice-chair Upcoming meetings are: 9 Nov - 13 Nov New Haven (Yale) 8 Feb - 12 Feb Ft. Lauderdale (Harris) Q21) How can I read my VAX binary data somewhere else? Some vendors provide bulit in methods (DEC provides this via special options on the OPEN statement). Others provide library support (on SPARC products, with the SunPro compilers, checkout convert_external) In addition, Accerl8 provides a commerial tool. Contact: Robert Hickler Tel (303) 863-8088 Accelr8 Technology Email robert@accelr8.com for more information. The following three Q's and A's based on email from: ecmtwhk@ccu1.aukuni.ac.nz (Thomas Koenig) Q22) My F77 program compiled ok on a , but gives me heaps of syntax errors on a . What's wrong? Most likely, the program was written with a line length greater than 72. If your compiler supports it, turn on the option for greater line length (e.g. -e is not uncommon) ; otherwise, split up the lines by hand, or via one of those pretty-printers/restructing tools mentioned above. Q23) My F77 program ran ok on a , but on a it just gives me strange results. What's wrong? There are different reasons why this could be. Possibly, your program violates the standard in some way which is not caught by the compiler on . Many programs rely on variables to be initialized to zero when a subroutine or function is first called, and also on the retention of values between calls. While many compilers (VAX/VMX and IBM VS, for example) exhibit this behaviour, compilers on newer architectures often fill variables with garbage on each new function call. This is permitted, according to the standard. Initialize variables explicitly, and put the variables which you need to keep across function calls into SAVE statements. Your compiler may have an option to trap unititialized variables; use that to find the trouble spots. Alternatively, compile using a flag which forces static allocation of all variables. Another problem might be that the accuracy of REAL and DOUBLE PRECISION differs between different platforms; that can cause roundoff error to wipe out your results or your program to go into endless loops. Yet another (and much more subtle) problem can occur if a lot of formatted I/O is employed. The conversion from internal to external representations can introduce very significant errors; much worse on some platforms than others (doing correctly rounded base conversion is expensive). There are, of course, lots of other possibilities, these are just a starting point. Q24) Is there a WEB for Fortran (and what is web anyway)? fweb is available via anonymous ftp; use archie(above) to determine current location(s). As for the more general question, what is WEB .... WEB is a "literate" programming system created by Knuth (he of reference book fame, and TeX fame, and etc.). It permits the programmer to write code and documentation together. WEB takes the "high level code and documentation" and creates real publishable documentation (using TeX or LaTeX) and compilable code. WEB versions for many languages are available. Q25) Where can I find coded BLAS (and what are coded BLAS?) The BLAS (basic linear algebra software) comes in several flavors: BLAS-1, -2, and -3. These can be described as scalar, vector and matrix-matrix levels. "Coded" BLAS are either hand coded in assembler, or at least tweaked for a given machine. Some vendors provide these, some are provided on the net (see archie) and some are marketed by various commercial organizations. In addition, it should be noted that BLAS-3 is very amenable to parallel processing. Done cleverly, this could be done by a network of processors over a net. DSS markets just such an implementation. Contact: Mike Boucher Dakota Scientific Software, Inc. 501 East Saint Joseph Street Rapid City, SD 57701-3995 fax: (605) 394-1256 scisoft@well.sf.ca.us for more information about their implementation. Q26) How common is DO ... END DO It is very common; and of course is part of Fortran 90. Compilers claimed to *not* support it (much shorter list this way) 1) Salford ftn77/Primos version 2) Prime f77 compiler 3) Microsoft Fortran for CP/M 8080/Z80 machines 4) Fujitsu VPxxx UXP/M compiler Q27) Where can I learn more about the history of Fortran? From: metcalf@apofort.cern.ch (Michael Metcalf ) The history of Fortran is documented in: Annals of History of Computing, 6, 1, January, 1984 (whole issue). Programming Systems and Languages (S. Rosen ed.), McGraw Hill, 1967, pp 29-47 (this is Backus's original paper). History of Prorammining Languages (R.L. Wexelblat ed.), Academic Press, 1981, pp 25-74. A summary appears in vol. 5 of the Encyclopedia of Science and Technology, Academic Press, 1986, under 'Fortran'. and in Chapter 1 of Fortran 90 Explained (Oxford, 1990). Q28) Fortran text editors? Thomas Koenig, ig25@rz.uni-karlsruhe.de, ig25@dkauni2.bitnet If you use EMACS, you can use M-x fortran-mode. An extended version, which includes the fortran-auto-fill-mode minor mode, is available via anonymous ftp from hallc1.cebaf.gov [129.57.32.62] in the directory /emacs. More generally, Unipress emacs has a Fortran mode. There are XEDIT, BRIEF, TPU and EDT clones available on many common platforms these days. -- ---------------------------------------------------------------- Keith H. Bierman keith.bierman@Sun.COM| khb@chiba.Eng.Sun.COM SunPro 2550 Garcia MTV 12-40 | (415 336 2648) fax 964 0946 Mountain View, CA 94043 Copyright 1993