1.60-b2

	- [BUG] writing equatorial maps had wrong reference position in FITS
	(it was inverted by mistake). This bug was introduced only in the last
	beta relase with the projection update... 	

	- [BUG] imagetool's '-clip=' option was slightly misbehaving for 
	equatorial maps following the projection update. This was only 
	introduced in the 1.60-b1 release, and affects no earlier releases.

1.60-b1
	
	- [NEW] Filenames and paths can now contain '~', '~user' or environment
	variables (enclosed in {}'s when in the config file). This practically
	affects configuration files only, since on the command line the shell
	already expanded these. Say you have the environment variable

		DATADIR = "mydata/sharc2"
	
	defined in the shell. Then, you could already before reduce data with
	
		> crush -path=~/$DATADIR [...]

	What's new is that now you can equivalently place

		RAW_DATA_PATH = ~/{$DATADIR} 

	into your configuration file. Notice that the reference to the 
	environment variable DATADIR is now enclosed in curly brackets!

	- [NEW] 'coadd' has been entirely rewritten! On the one hand it is
	made simpler, and on the other hand also more reliable and generally
	better. The new coaddition uses the proper spherical projection
	(which can be selected via the -projection= option, see below for more)
	The following options are now supported:

		-projection=   	Select the desired spherical projection scheme
				Supporting: SIN, TAN, ZEA, SFL, AIT, MER, CAR
	
		-weighting	Iteratively derive weights for the component
				maps.

		-scaling	Fit optimal relative scaling of components
				(useful if there are peak variations.)

		-offsets	Fit optimal zero level offsets.
				(useful for coadding very extended sources.)

		-noise=		'data' or 'image'. Define whether to use
				noise weights that are derived from the time-
				streams, or directly from the maps.

	All other operations (e.g. explicit scaling or offseting, clipping etc.)
	before or after coadding should be done via 'imagetool'.

	- [NEW] At last CRUSH provides proper map projections as opposed
	to the fudged approximation that was used pior. The difference is
	unnoticable for the typical small SHARC-2 field size (a few arc 
	minutes accross), but becomes more pronounced for large maps where 
	the the 'curvature' of the mapped area is noticable.
	Map projections can be selected via the MAP_PROJECTION key or the
	corresponding -projection= command-line option. 

	The argument is either the full name (in quotes if it contains white 
	spaces, or with the white spaces and dashes scrapped) of the projection
 	or its 3-letter FITS abbreviation. The following projections are 
	currently available:

		SIN -- Slant Orthographic
		TAN -- Gnomonic
		ZEA -- Zenithal Equal Area	
	 	CAR -- Plate-Carree
		MER -- Mercator
		SFL -- Sanson-Flamsteed
		AIT -- Hammer-Aitoff

	The default is TAN (Gnomonic), as it is widely supported by FITS
	viewers. The implementation of the projections follows the prescription
	by Calabretta & Greisen (2002).

	- [NEW] New features in extended structure filter:

		* Added filtering via FFT
		* Can specify a S/N level that is blanking the filter
		* Masked circular regions are also excluded from the
		  filter (see mask.example for what a mask looks like).

	  In 'crush', the type of filter can be selected via the new key
	  EXTENEDED_FILTER_TYPE, which can have values either "fft" or
	  "convolution". The S/N blanking is adjusted via the 
	  EXTENDED_FILTER_BLANKING key. The corresponding options work
	  somewhat differently. the new -fftFilter= key works the same
	  way as the prior -extFilter= key, specifying the filter scale,
	  but selects the FFT mode. The blanking is defined via the 
	  -extBlank= key, which takes either the S/N value or "off". E.g,
	  the keys:
		
		EXTENDED_FILTER_SCALE = 45.0
		EXTENDED_FILTER_TYPE = fft
		EXTENDED_FILTER_BLANKING = 6.0

	  is equivalent to
	
		crush [...] -fftFilter=45.0 -extBlank=6.0 [...]

	  In 'imagetool' and 'show', the -extFilter and -fftFilter keys
	  are work similarly, except that the blanking value can be defined
	  after the filter FWHM separated by a comma. E.g.

		imagetool -fftFilter=45.0,6.0

	  will filter structures over 45", while leaving those parts of the
	  map that are over 6-sigma unfiltered.

	- [BUG] Weights were not correctly normalized when regridding (-regrid)
	leading to wrong rms and S/N estimates. Sorry! At least, it is now 
	fixed.

	- [NEW] Added '-s2nclip=' option to 'imagetool' and hence also to 
	'show', allowing to clip maps below the specified signal-to-noise 
	value.

	- [NEW] crush image manipulation tools can now read and manipulate BoA
	images.

	- [UPDATE] Deconvolution (see imagetool's '-deconvolve=' option) has
	been improved. It now gives a cleaner results and more reliable
	noise estimates on the deconvolved images.

	- [BUG] Flux correction could result in NaN's if REG_SIZE key was
	not present in FITS. This has been fixed.

	- [FEATURE] 'show' now measures aperture fluxes correctly, for 
	compatible but non-SHARC2 images as well! (e.g. LABOCA, ASZCA)

	- [NEW] removed -epoch= option. This is really not necessary as
	the coordinate epochs are correctly recorded in the scan files.

	- [NEW] 'show' and 'imagetool' are made more verbose on map operations.

	- [UPDATE] Improved on flux correction scheme. The scheme can 
	simultaneously treat extended filters and regional correlation 
	modeling, and takes map clipping and filter blanking into account 
	when calculating corrections.

	- Minor tweaks: 

	   * unsmoothed map has effective smoothing FWHM of resolution/1.13 
	     This reflects that a Gaussian beam area is ~1.13 FWHM^2, and
	     pixelization is essentially a smoothing by the pixel area.

	   * flux corrections are applied to maps during reduction.

	   * Extended filter is now applied to the full map not the scan maps.
	     This has the advantage of better blanking of faint but 
	     significant structures, but comes as the cost of a little 
	     weaker scan weighting... 

	   * Cleaned up source fitting. Should be a little more robust...

	- Small tweaks to 'histogram' tool. Rounded bin boundaries and cleaned
	output.

1.52-1 (unreleased)

	- No longe levelling scan maps. It's not necessary.

	- Added map sanitization to make sure clipped data is not synched
	with the timestreams. clipping functions are also simpler as they only 
	raise flags.

	- [NEW] Negative clipping is changed to signal-to-noise clipping 
	rather than flux clipping relative to map peak. Changed configuration 
	files accordingly and updated README. S2N clipping is now default for
	regular and '-faint' objects for 2 iterations but it is disabled for
	'-deep' reductions. Most importantly one should remember not to use 
	the -correct feature together with s2n clipping. Thus 
	MAP_CUTOFF_LEVEL = 5.0 or -cutoff=5.0 (default) specifies to clip
	preliminary source maps at 5-sigma.

	- Changed map post-processing order.

	- [NEW] -extFilter now excludes regions in masks from filtering. This 
	allows filtering extended structures around bright compact sources 
	without producing the negative dips around. Mask can be produced either 
	manually (see mask.example) or by selecting circular regions in the
	'show' tool with the mouse (left-click center, drag, left-click again 
	for radius).

	- [BUG] Fixed error with parsing negative angles in D:M:S format, and 
	times in H:M:S format. This affected the reading of mask files when 
	one of the coordinates happened to be negative.

	- [NEW] Change flux correction algorithm to account for filtering of 
	large scales via -extFilter, so now fluxes are approximately correct 
	even if such explicit filtering was applied. Exception is the 
	simultaneous use of regional or FFT filtering used together with 
	-extFilter. The correction is likely to get confused in this case. 
	Advice is to stick to one spacial filtering method at a time...

	- [BUG] Small errors with the multiple point source extraction method 
	are fixed. (These were corrected for large scale filtering effects even
	when such filter was not applied).	

	- [NEW] Improvements to account for filtering of extended structure in 
	the flux correction scheme. Now approximately correct fluxes are read 
	from maps which had extFilter=X applied to them, provided the 
	-correct=X option was approproately used.

	- ! Slight issues with auto-ordering pipelines. Models which are tied
	together (e.g. correlated and gain) can switch order under certain 
	conditions...
