mab@ukc.ac.uk (M.A.Broom) (02/22/91)
Does anybody know of a good Fortran compiler for a
Unix Operating system ?
We would like it to be able to check the following:
a) extensions to the Fortran 77 standard;
b) check array bounds are within defined range;
c) check the types of arguments passed between program
units are the same and the number of arguments
passed is correct;
d) undefined variables are left as undefined and not set to
zero or any other predefined initial state;
e) check varaibles have been declared before used.
At the moment we have a DEC 8800 system running VMS and we
are using the University of Waterloo WATFOR-77 compiler which
does all the above.
If anybody knows of a good fortran compiler for a unix opearting
system that does at least items (c) and (d) above could they mail me.
Thanks in advance.
Martin Broom mab@ukc.uk.ac
Computing Lab
University of Kent
Canterbury
Kent
Englandkhb@chiba.Eng.Sun.COM (Keith Bierman fpgroup) (02/23/91)
In article <6939@harrier.ukc.ac.uk> mab@ukc.ac.uk (M.A.Broom) writes: Does anybody know of a good Fortran compiler for a Unix Operating system ? We would like it to be able to check the following: a) extensions to the Fortran 77 standard; many claim this, few do it really well (vms fortran goofs in some places, and it is very good at this sort of thing). IBM's xlf is very strict. In general, I strongly advise the use of a static analysis tool (a review of one in particular "attached"). b) check array bounds are within defined range; -C is supported by most unix f77 compilers. it provides runtime checks c) check the types of arguments passed between program units are the same and the number of arguments passed is correct; xlf does this. static analyzers do it. d) undefined variables are left as undefined and not set to zero or any other predefined initial state; static analyzers do this. e) check varaibles have been declared before used. ditto Subject: fortran-lint (fortran tool(s)) --text follows this line-- Intro ----------------- From time to time folks ask: 1) Can I get numbered listings (sure awk "{i=i+1; print i":" $0}" etc.) 2) Can I get Cross references (sure use source browser) 3) Why is my code broken (hours later, ah is this a O or an 0?) 4) Why doesn't cxflow work for f77 ? There are a variety of tools, public domain, site specific and commerical which address these needs. One of the groups which sells such a product was kind enough to send it to me to review. Top Line ----------- The product solves far more problems than it creates. Certainly handy to have around. Has a few rough edges, but very, very useful. Details --------------- The vendor is IPT 1096 East Meadow Circle, Palo Alto, CA 94303 415-494-7500 *** legalism: PRL of the UK claims "flint" is a name trademarked by them. In the text of this review I use "flint" as the invocation name of the IPT product (because it is). I have no reason to doubt PRL's claim so this is acknowledgement that PRL claims trademark protection of the name. If possible, I will review their flint at a later date. *** end aside The most obvious thing for a "lint" program to do, is to catch real and likely coding errors (such as mismatched argument lists, variables used before defined, violations of standards, etc.). IPT fortran-lint does a crediable job of this. Several different levels of checking are available. A simple flint foo.f will catch things like USAGE WARNING- local variables set but never referenced: NUMSIT (Line 2199) USAGE ERROR- local variables referenced but never set: NFLVPR (Line 306) clearly the first is just bad juju (old code not maintained well, or a possible typo) the latter is a sure sign that the code may fail with the optimizer is turned up, or the code moved to another machine, or recursion is enabled, or ... SYNTAX WARNING- branch to label 29 from outside do loop. I would have made this more than a warning.... as the code really looks like go to 29 do 30 im=1,nm 29 continue hs(nslice+im)=0.0 30 continue which is not standard complying ... (so I'd call it an error, especially with the -a switch see below). But the key thing, is that it is identified ... which makes it likely to be caught. There is the added flint -a foo.f > common > * cstop , dcycst, dmpint, dmplst, dmptime, dradlos, dtld , > ^ flux.f:FLUX line 113: ANSI EXTENSION- symbol name longer than 6 characters. Which, as can easily be seen, adds strict ANSI checking. There are a variety of flags, and tons of messages. Far more than I am willing to clutter up a review with. We will only hit _some_ of the highlights. flint -g foo.f | *.f | `cat mylistoffiles` enables checking of interfaces. Checking is done for correct type (real, double, logical*n, etc.) and size (real foo(100) vs real foo) for storing into a constant. Consider the following silly codelet program bork real*8 x,y call foo(x,1.0d0) print*,x end subroutine foo(x,y) real*8 x(100) real y do i = 1, n y = x(i)**2+y enddo return end flint -g silly.f produces FORTRAN-lint Rev 2.61 7-Mar-90 22:54:11 Options: -g Directory (current) bork2.f ******************************************************************************* Program BORK File bork2.f Line 1 > call foo(x,1.0d0) > ^ bork2.f:BORK line 4: INTERFACE ERROR- R*8 variable passed to dummy arg which is a R*8 array. <there is a way to set things up so this can get past sans nattering ... as many large codes do this on purpose> > call foo(x,1.0d0) > ^ bork2.f:BORK line 4: INTERFACE ERROR- R*8 actual arg passed to a R*4 dummy arg. <<this is likely to be OK on DEC, wrong everywhere else>> > call foo(x,1.0d0) > ^ bork2.f:BORK line 4: INTERFACE ERROR- constant is changed by subprogram. <<this is likely to be very confusing>> USAGE ERROR- local variables referenced but never set: X (Line 4) ******************************************************************************* Subroutine FOO File bork2.f Line 7 USAGE ERROR- local variables referenced but never set: N (Line 11) ******************************************************************************* Global checking: --- No additional global inconsistencies found --- This functionality would be enough to warrent having this puppy handy. But like a late night commerical (which I should be home watching ;>) there's more! There is a relatively comprehensive xref. The layout is a bit space wasteful, here is a sample: FORTRAN-lint (symbol table) 5-Mar-90 16:03:32 ******** SYMBOL TABLE ******** *** Program: SPICE : defined at line 2 of file spice.f Calls- cputim.f:CPUTIM, elapse.f:ELAPSE, getcje.f:GETCJE, tsec.f:TSEC, setmem.f:SETMEM, zero8.f:ZERO8, readin.f:READIN, errchk.f:ERRCHK, setup.f:SETUP, alter.f:ALTER, tmpupd.f:TMPUPD, dctran.f:DCTRAN, ovtpvt.f:OVTPVT, dcop.f:DCOP, acan.f:ACAN, title.f:TITLE *** Subroutines: ACAN : defined at line 14 of file acan.f Args- none Calls- tsec.f:TSEC, getm8.f:GETM8, getm16.f:GETM16, dinit.f:DINIT, crunch.f:CRUNCH, getcje.f:GETCJE, acload.f:ACLOAD, acdcmp.f:ACDCMP, acsol.f:ACSOL, extmem.f:EXTMEM, copy16.f:COPY16, noise.f:NOISE, disto.f:DISTO, clrmem.f:CLRMEM Called by- spice.f:SPICE ... CMEYER : defined at line 1 of file cmeyer.f Args- (R*8 R, R*8 R, R*8 R, R*8 R, R*8 R, R*8 R, R*8 R, R*8 R, R*8 R, R*8 R, R*8 R, R*8 S, R*8 S, R*8 S, R*8 S, R*8 S, R*8 S) Called by- mosfet.f:MOSFET .... note that R*8 R means dp, referenced, S means Set. .... *** Vars/Arrays: A (adj) : R*8 : local in (shlsrt.f:SHLSRT) is Ref, Set, Dummy argument A (1) : CHAR*1 : local in (move.f:MOVE) is Set, Dummy argument A : R*8 : local in (moseq2.f:MOSEQ2) is Ref, Set in (scale.f:SCALE) is Ref, Set in (xxor.f:XXOR) is Ref, Dummy argument .... this provides handy information. Source browser is more fun, and almost certainly the better tool. But there are times where this sort of report is exactly what management ordered.... There are summary statistics FORTRAN-lint (statistics) 5-Mar-90 16:03:32 >>> Statistics: Number of source files: 130 Source files: 18556 lines, 561873 bytes ( 14% comments, 86% code ) Include files: 0 lines, 0 bytes Total parsed: 18556 lines, 561873 bytes ( 14% comments, 86% code ) Total subprograms: 130 Subroutines: 121 Functions: 8 Program: 1 Block Data: 0 Total messages: 652 Errors Warnings FYIs -------- -------- -------- Syntax: 0 0 102 Interface: 426 1 4 Data usage: 17 58 44 Implicit typing: <supp> Portability issues: 0 ANSI extensions: <supp> supp means suppressed. It seems strange to note interface errors but no portability issues. The defintion of portability issue is possibly debatable. ---- Call graph tree, with options to minimze output. {means missing} FORTRAN-lint (call tree) 5-Mar-90 16:03:32 Call tree: This is a primary tree starting at the program 'SPICE' SPICE-+-CPUTIM (1)--{ETIME} +-ELAPSE--{TIME} +-GETCJE (2)--TSEC (3)--CPUTIM see 1 +-TSEC see 3 +-SETMEM-+-LOCF | +-NXTEVN | +-NXTMEM | +-ERRMEM (4)--DMPMEM (5)--LOCF | +-MEMORY +-ZERO8 +-READIN-+-TSEC see 3 | +-ZERO4 | +-GETLIN (6)--COPY8 (7)--LOCF | +-COPY8 see 7 | +-GETM4 (8)--GETMX (9)-+-MEMPTR (10)--LOCF | | +-ERRMEM see 4 | | +-NXTEVN | | +-COMPRS (11)-+-NXTEVN | | | +-COPY4 (12)--LOCF | | +-NXTMEM | | +-MEMORY | | +-COPY4 see 12 | | +-LOCF | | +-MEMADJ (13)-+-NXTMEM | | +-COMPRS see 11 | | +-COPY4 see 12 | | +-MEMORY ..... This is a detached tree starting at the subprogram 'ZERO16' ZERO16 ..... The detached tree means code left in, which is called from nowhere. Probably best to remove, though there are times when someone does this on purpose.... I have found a fair number of false positives, and there are a couple of modest bugs. I will be doing more testing, and reporting bugs to the authors. The documentation does not do the product justice, and it could use some manpages or other online reference facility. I don't know pricing. There is a copy protection scheme (hostid based at the moment, but a floating server is being implemented RealSoonNow). There seem to be quantity discounts. ...sun internal remarks deleted... enhancement-wise, this fortran-lint from IPT (or equivalents from elsewhere) will be a great boon to those who wrestle with Fortran codes, and sometimes lose. There are other vendors with similar (and perhaps even superior) tools, but I haven't reviewed them personally .... so for the nonce this is my new answer to questions (1)-(4) from the intro. Cheers all. Keith H. Bierman |*My thoughts are my own. !! kbierman@Eng.Sun.COM It's Not My Fault | MTS --Only my work belongs to Sun* kbierman%eng@sun.com I Voted for Bill & | Advanced Languages/Floating Point Group Opus | "When the going gets Weird .. the Weird turn PRO" "There is NO defense against the attack of the KILLER MICROS!" Eugene Brooks Postscript: Some competing products: FORWARN Quibus Enterprises: Seems similar, from a promo blurb reading. Probably has slightly uglier reporting (one assumes the promo is as good as it gets). 217 356 8876. Primary focus is the PC market, but say unix versions are available. $1200 would appear to be the price, no discount structure info available. Also avaliable from them are "Fortran development tools" reformatter, simple spagetti to if-then-else. Flint(tm) Programming Research ltd: Seems like a much more powerful tool (from the promo lit) includes complexity metrics, X11 interface and all sorts of advisories about what constitutes good portable code (beyond standard(s) conformance). FAX (44) 01 336 1151 voice (44) 1 942 9242 also 8701 Bedford Euless Rd. Suite 520 Hurst TX 76053 USA ph: 817-589-0949, fax: 817-595-4611 FORCHECK Leiden University Box 9604 2300 RC Leiden The Netherlands 31-71-276804 Claims similar to IPT's product. Is said to be available on Suns and many other platforms. I have no personal experience with it. FLOPPY/FLOW From CERN. FLOPPY was posted to comp.sources.misc. Sun employees can get it via the usual internal anon ftp site. FLOW was not posted (and is the 'good part'). Related tools of interest: FOR_STRUCT Cobolt-Blue 2940 Union Ave Suite C San Jose CA 95124 408 723 0474. This is a "reverse engineering tool" to use the COBOL lingo, viz. makes old code more readable and maintainable. I have not used it myself. SPAG I have reviewed SPAG, short overview is that SPAG is quite handy ... alas not very unixlike (fixable with some shell programming). (Sun employees note: Full review is in the file called SPAG in the same directory as this review). I like the results, and recommend this tool with the caveat about non-unix interface... US distributor: OTG voice 717 222 9100 fax 717 229 9103 Authors: Polyhedron Software Ltd Magdalen House 98 Abingdon Road Standlake Witney Oxon OX8 7RN Tel 0865 300 579 -- ---------------------------------------------------------------- Keith H. Bierman kbierman@Eng.Sun.COM | khb@chiba.Eng.Sun.COM SMI 2550 Garcia 12-33 | (415 336 2648) Mountain View, CA 94043