iraf.net

NUMBER:	513
MODULE:	icfit (apextract, curfit, ...)
SYSTEM:	V2.12-V2.12.1
DATE:	Mon Nov 18 16:11:42 MST 2002
FROM:	valdes

BUG:	When a single fitting (sample) region is defined and binning is used
	(|naverage| > 1) such that only a single binned fitting point
	results, then incorrect results or a segmentation error may occur.

	A case where this occurs is in apall/apsum when a single background
	region is defined and a large binning value is set to use the
	median or average of the background sample regions.  An inobvious
	situation resulting in this is when there are actually two
	background regions on either side of an aperture which are defined
	relative to the aperture.  If the aperture is defined or moves too
	close to the edge of the image such that one of the background
	regions goes entirely off the image, then this error can be
	triggered.

	This problem behavior was introduced in V2.12 when the stability of
	the fit was improved by only fitting only over the range of the data
	selected in by the sample region(s) rather than over the entire
	data set.

	The only workaround is to avoid this case.  For apall/apsum one
	needs to check for a background region going off the edge or be
	sure to use |naverage| smaller than the width of the background
	regions to insure at least two fitting points.

STATUS:	The case of a single fitting point in binned data will be handled
	properly in the next release.