RELEASE NOTES 

		spatstat 1.11-0

		12 January 2007
	
OVERVIEW

	Internals of the package are being reorganised,
	in preparation for a major upgrade.

	Format of ppm objects changed.

	More stringent checking of function arguments.

	Improved handling of pixellation effects.

	Extensions to variance-covariance calculations for fitted models.

	4 bug fixes.

COMPATIBILITY

	The internal representation of "ppm" objects has changed.

	Point process models that were fitted using ppm() 
	in older versions of spatstat are in a format that is now out-of-date.

	These old objects are still handled correctly by spatstat 1.11-0,
	as far as we are able.

	We recommend that these old objects be updated to the current version
	of spatstat, using `update.ppm', or of course by re-running
	your original code.

	The internal representation of "interact" objects has also changed.
	These objects are created by the functions Poisson(), Strauss(),
	DiggleGratton(), Geyer(), LennardJones(), MultiStrauss(), 
	MultiStraussHard(), Ord(), OrdThresh(), PairPiece(), Pairwise(), 
	Saturated(), StraussHard() and Softcore().
	Similar comments apply.

	The best advice is to ** save only R scripts ** that generate
	fitted models, rather than saving the fitted models themselves.

NEW FUNCTIONS
	
	[.splitppp
	[<-.splitppp
	summary.splitppp
		Methods for the class "splitppp".

	[.listof
	[<-.listof
	summary.listof
		Methods for the class "listof"

	as.owin.ppm
		Method for "as.owin"

IMPROVEMENTS/ENHANCEMENTS

	Kinhom
		Improved handling of NA's and pixellation effects. 
		If 'lambda' is a pixel image which does not completely cover
		the observation window of the point pattern X, then 
		X will first be trimmed to a smaller region where the values
		of lambda are defined. A warning is issued.

	intersect.owin
	union.owin
		Better handling of pixel rasters. 
  		 (a) If one of the windows is already a binary mask, 
		the default pixel raster is taken from this window.
		 (b) These functions now have a "..." argument which 
		gives complete control over the pixel raster dimensions. 

	ppp
	as.ppp
		These functions now check that the (x,y) coordinate points
		all lie inside the window specified.
		Checking may be switched off by setting check=FALSE.

	ppm
		Now applies more stringent checks on the validity of
		its arguments `trend' and `interaction'.

	ppm
	summary.ppm
		These now handle the case where covariates = list()
		or `covariates' is a data frame with zero columns.
		
	qqplot.ppm
		Errors that occur during simulation are now trapped
		instead of causing a crash. 
		New argument 'maxerr' controls maximum number of such errors
		that will be tolerated.
		New default value for 'periodic'.

	plot.listof
	plot.splitppp
		These functions have additional arguments 'nrows' and 'ncols'
		to control the plot layout.

	rpoispp
	rNeymanScott
	rMatClust
	rThomas
	rSSI
		These functions now check their arguments more stringently.

	Kest
		New option: correction="none" computes an estimate of
		the K function without edge correction. This is a heavily
		biased estimate and should not be used for data analysis.
		It is useful for demonstration/instruction purposes.

	vcov.ppm
		Now computes Monte Carlo estimate of Fisher information
		and variance-covariance matrix for any model fitted
		by Huang-Ogata method (by calling ppm with method="ho").

	vcov.ppm
		New argument 'gamaction' to control handling of 
		models that were fitted using gam().

`INVISIBLE' CHANGES

	This release contains numerous changes to the internal code
	for point process models (ppm objects). These changes are in 
	preparation for a future upgrade to spatstat. Most of the changes
	are not currently visible.

BUGS

	[1] bounding.box
		The bounding box was incorrect for a pixel image
		and for a window of type 'mask'. The bounding box
		was half a pixel too small on each side.
		Fixed.

	[2] dilate.owin
		Due to discretisation errors, 
		dilate(w, r) did not always contain w, for r >= 0,
		This has now been modified so that 
		dilate(w, r) always contains w, and 
		dilate(w, 0) returns w.

	[3] lurking, diagnose.ppm
		Calculation of variance failed (with a message about
		`non-conformable arrays') when the fitted model 
		had more than one covariate. 
		Fixed.

	[4] vcov.ppm, lurking
		These functions crashed when applied to a point process model
		that was fitted using gam (i.e. by calling ppm with 
		use.gam=TRUE).
		Fixed.