ABOUT THE TEST-VECTOR GENERATORS

This directory contains the programs that generate the test vectors
that are used in the test environment.  Each program generates at
least two functions (float and double precision).  Output consists
of a file for each function covered whose name is <function>_vec.c.
In MSDOS, this name is shortened as needed to fit in the format of
an eight-character file name.

Name of		Functions
Generator	 Covered
---------	---------
gacosvec.cpp	acos	acosf
gacoshvec.cpp	acosh	acoshf
gasinvec.cpp	asin	asinf
gasinhvec.cpp	asinh	asinhf
gatanvec.cpp	atan	atanf
gatan2vec.cpp	atan2	atan2f
gatanhvec.cpp	atanh	atanhf
gcoshvec.cpp	cosh	coshf
gcosvec.cpp	cos	cosf
gerfvec.cpp	erf	erff
		erfc	erfcf
gexpvec.cpp	exp	expf
gfmodvec.cpp	fmod	fmodf
gfrexpvec.cpp	frexp	frexpf
ggammavec.cpp	gamma	gammaf
ghypotvec.cpp	hypot	hypotf
gjy01vec.cpp	j0	j0f	j1	j1f
		y0	y0f	y1	y1f
gjynvec.cpp	jn	jnf	yn	ynf
gldexpvec.cpp	ldexp	ldexpf
glog10vec.cpp	log10	log10f
glog1pvec.cpp	log1p	log1pf
glog2vec.cpp	log2	log2f
glogvec.cpp	log	logf
gmiscvec.cpp	ceil	ceilf
		fabs	fabsf
		floor	floorf
gmodfvec.cpp	modf	modff
gpowvec.cpp	pow	powf
gsinhvec.cpp	sinh	sinhf
gsinvec.cpp	sin	sinf
gsqrtvec.cpp	sqrt	sqrtf
gtanhvec.cpp	tanh	tanhf
gtanvec.cpp	tan	tanf


The principle code files for each of the generators consist of a *.cpp
file and a header file, genmathv.h.  The code for each generator is built
quite like each of its brethren:

    There are two generating subroutines - one generates test vectors
    for double precision functions, the other for float precision ones.
    Each generating subroutine does the following:

	1. Opens a disk file for output
	2. Writes the first two lines of the vector file
	3. Generates vectors, writes to disk

    The main function is located at the very bottom of each program
    file.

Most of the generators use functions in the extra-precision library to
produce base values.  The extra-precision functions, housed in directory
tests/cygnus/tgen/qfloat, are defined in header file qfloat.h.  This file
is a C++ wrapper file that defines inline functions for extra-precision
arithmetic.

To compile a generator, test.h and qfloat.h must be in their respective
directories and the extra-precision library, qfloatx.a, must be located
in the qfloat directory.  The makefile in this (the tgen) directory is
fairly straightforward.

K.B. Williams
Kbwms@aol.com