V10/cmd/f2c/doc/st

.EQ
delim $$
define f2c % "f\|2c" %
define F2c % "F\^2c" %
define libF77 % "libF77" %
define libI77 % "libI77" %
define LibF77 % "LibF77" %
define LibI77 % "LibI77" %
.EN
.nr PS 24
.nr VS 30
.nr PO .5i
.nr LL 7.5i
.LP
.ds CH "\s8\(hy \\n(PN \(hy\s0
.ce
$f2c$ (Stu Feldman, Mark Maimone, Norm Schryer)

.nf
.ta .5i .8i 1.1i
Convert Fortran 77 to C
.sp . 5
	\(**	Portably mix C and Fortran
	\(**	Use Fortran source in C environments

.fi
\s20From cs.purdue.edu!peb Mon Jan 29 15:35:20 EST 1990\s0
.sp .5
BTW, I think f2c has the potential to be the best idea since sliced bread.
I have tons of Fortran code, but my collaborators use C, now
we can stop arguing about what language to use and get some work done.
.sp 1.5
.nf
Seek binary compatibility with VAX f77
		\(bu	big help in testing
		\(bu	helped find bugs in $f2c$, $lcc$, $nCC$

Useful test cases:
		\(bu NBS tests
		\(bu PORT3
.bp
Netlib's "execute f2c" taught us much
			(starting 19 Nov. 1989):

.ce
\s22\f(CW{echo execute f2c; cat *.f} | mail netlib\fR\s0

		\(bu	bugs
		\(bu	\f(CW-z\fR (no implicit \f(CWdouble complex\fR)
		\(bu	longer names
		\(bu	better error messages (as late as today)
		\(bu	args on main
		\(bu	\f(CWblock data\fR initializing vars not in \f(CWcommon\fR
		\(bu	\f(CWdo ... enddo\fR
		\(bu	\f(CW!\fR comments
		\(bu	\f(CWnamelist\fR


	\(bx COMMON

	\(bx constants (preserved until operated on)

	\(bx prototypes
.bp
Namelist example:
.ft CW
.sp p.5
		integer m(8)
		real x(10,10)
		namelist /xx/ m, x
		\&. . .
		read(*,xx)

\fRcould read\fP

		&xx x(1,1) = 2,
			x(1:3,8:10:2) = 1,2,3,4,5,6
			m(7:8) = 9,10/

.ft R
Recent lesson: many vendors expect non-8x syntax:

.EQ
delim off
.EN
.ft CW
		$xx x(1,1) = 2,
			x(1:3,8:10:2) = 1,2,3,4,5,6
			m(7:8) = 9,10 $end
.ft R
.EQ
delim $$
.EN
.bp
Prototypes:
.sp .5
		\f(CWf2c -P foo.f\fR
.sp .5
creates both \f(CWfoo.c\fR and \f(CWfoo.P\fR .

Examples of PORT3 prototypes:

.ft CW
	extern C_f cdotc_(complex *ret_val,
			integer *n, complex *dx,
			integer *incx, complex *dy,
			integer *incy);

	extern C_f mullr_(complex *ret_val,
			C_fp f, complex *z1,
			complex *z2, complex *z3,
			real *epsz, real *epsf,
			integer *maxitr, integer *iter);


f2c.h \fRcontains\f(CW typedef\fRs.
.bp
Special comments for \f(CWcommon\fR lengths (read by $f2c$):

.ft CW
	/* comlen a0desp_ 4 */
	/* comlen a10der_ 36 */
	/* comlen a10dey_ 480 */
.ft R


Special comments for references:
.ft CW
.ps 16
	/*:ref: odes1_ 10 11 200 4 3 4 4 4 208 4 210 8 8 */
	/*:ref: l4ags_ 10 14 4 3 3 3 210 3 4 4 4 3 3 3 3 4 */
	/* Rerunning f2c -P may change prototypes or declarations. */
.ps 24
.ft R


Untyped \f(CWexternal\fRs
			\(=> need to update prototypes:

.EQ
delim off
.EN
.ft CW
	while :; do
			f2c -Ps f.[fP]
			case $? in 4) ;; *) break;; esac
			done
.ft R

\(bx Several PORT3 bugs found.
.bp
Can't always infer types correctly:

.ft CW
	subroutine grok(f,g,h)
	external f, g, h
	logical g
	call foo(1,g)
	call foo(2,f)
	call zap(1,f)
	call zap(2,h)
	call zot(f(3))
	end
.ft R

Infer \f(CWf_\fR and \f(CWh_ to be \f(CWlogical\fR,
then find \f(CWf_\fR must be \f(CWreal\fR.
\f(CW-!it\fR turns such inferences off.
.bp
.ft CW
	subroutine zap(n,f)
	external f
	if (n .le. 1) call zap1(f)
	if (n .ge. 2) call zap2(f)
	end
.ft R




9 Jan. 1990 \- 30 Jan. 1990 (13:00):


	210 requests from 184 people
		for ``\f(CWall from f2c/src\fR''
			9 copies today


	695 requests from 627 people
		for ``\f(CWindex from f2c\fR''
			22 copies today