fft1d.par
t_fft1d.x
	Added "interval" and "angular" parameters. The first one is used
	to the sampling interval, and the second one is used to specify
	whether to use angular frecuencies in the output. With these
	parameters it is now possible to get the frecuency scale along
	with the transformation, in a two column output (PG 15/Nov/90).

t_fft1d.x
	Modified in order to be consistent with the changes in fft1d.x,
	specially with fft1_size() and fft1_get() (PG 16/Nov/90).

t_fft1d.x
	Increased the initial size of the FFT buffer from 128 to 512
	points (PG 16/Nov/90).

fft1d.par
t_fft1d.x
	Added "flip" parameter, and modified calls to fft1_init() in order
	to be consistent with changes on it. Also the code was modified so
	"flip" is turned on every time the sampling interval is defined.
	This is to prevent wrong scaling (PG 16/Nov/90).

t_fft1d.x
	Changed the image reading section so now the task process images
	that are not one dimensional (naxis > 1), but that have a only
	one line (PG 06/Jun/91).

fft1d.hlp
fft1d.par
t_fft1d.x
	Modified the behavior of the task with images. Now the task will
	get the sampling interval from the input image header by reading
	WPC and CDELT1 (in that order). The output image header will contain
	three new keywords: CRVAL1 with the starting frequency (always zero),
	CDELT1 with the frequency increment, and NYQUIST with the Nyquist
	frequency. The task will now output the square of the transformation
	instead of the modulus, so the "modulus" parameter was renamed as
	"power" to be consistent with this change. The help page was updated
	as well (PG 06/Aug/91).

fft11d.hlp
fft11d.par
t_fft1d.x
	Renamed the "power" parameter to "format", and changed it to a string
	parameter with three possible values: "plain" for the untouched output,
	"modulus" for the modulus of the transformation, and "power" for the
	square of the transformation (PG 07/Aug/91).

t_fft1d.x
	Changed the output image length to be the actual size of the FFT
	buffer, instead of the input image length (PG 03/Sep/91).

fft1d.par
t_fft1d.x
	Added two new parameters to specify the name of the image header
	keywords instead of using the fixed values "WPC", "CRDELT1", and
	"CRVAL1". Also, the "NYQUIST" keyword is no longer written in the
	output image header. The new keyword "FFT1DVAL" is now written when
	taking the direct transformation, in order to save the physical
	coordinate of the first pixel (old CRVAL1), and restore it when
	taking the inverse transformation (PG 03/Sep/91).
	
t_fft1d.x
	Modified the behavior of the "flip" parameter in such a way that
	now is not set automaticaly when the sampling interval is specified,
	either by using the "interval" parameter or the input image header
	keywords (PG 03/Sep/91).

t_fft1d.x
	Fixed two problems in fft1_image(). The first one was a bug related
	with the fact that fft1_init() was being called before the line
	length of the input image was retrived, thus initializing with the
	wrong buffer size. The second was a problem with always reading the
	image as complex. For some reason the data buffer returned by IMIO
	had garbage in at least one pixel value. The fix was to read as
	complex only complex images, and use a real read for all the others.
	The reason for this behavior is still unknown (PG 02/Dec/92).
