RELEASE NOTES 

		spatstat 1.13-1

		23 April 2008
	
	We thank Jeff Laake for several contributions.

OVERVIEW
	Important bug fix to rStrauss.
	Minor repairs to point process modelling, 
	segment clipping, and Metropolis-Hastings simulation code.
	Various minor improvements.

WARNINGS

	A bug has been detected and fixed in rStrauss(),
	the Perfect Simulation algorithm for the Strauss point process.

	The output of a single call to rStrauss() was correct,
	but repeated calls to rStrauss() were not stochastically independent.
	See bug [1] below. This has now been fixed. 

	Simulation studies that were conducted using rStrauss() 
	(in spatstat 1.13-0 and earlier) may have yielded incorrect results, 
	and should be repeated. 

	This does NOT affect simulations conducted using rmh(),
	qqplot.ppm(), envelope(), ppm(method="ho") etc. 
	unless you explicitly invoked the function rStrauss.

NEW FUNCTIONS

	stieltjes
		Computes Stieltjes integrals.

IMPROVEMENTS/ENHANCEMENTS
	
	rmh, envelope, qqplot.ppm
		Simulations of fitted point process models (ppm objects)
		now run slightly faster.

	model.matrix.ppm
		new argument 'keepNA' determines whether to retain or delete
		any rows of the model matrix that contain NA's. 

	clarkevans
		has a new option: correction="cdf" 

	ppm
		A warning message, which was printed if there were NA's
		in the covariate data, was rather confusing. 
		It has been clarified.


BUGS
	[1] rStrauss
		Random point patterns generated by successive calls 
		to rStrauss() were not stochastically independent.
		This occurred because R's random number generator seed
		was not correctly updated.
		Fixed.

	[2] vcov.ppm
		This function crashed when applied to certain 
		ppm objects that were fitted with use.gam=TRUE.
		(Detected by Jeff Laake).
		Fixed. 

	[3] "[.psp" 
		The segment clipping code generated an error
		if no segments required clipping. 
		(Detected by Jeff Laake).
		Fixed.

	[4] ppm
		When fitting a multitype point process model, 
		ppm issued an incorrect warning about duplicated points.
		Fixed.

	[5] rStrauss
		very occasionally terminated with an error message about
		memory allocation ("memory not mapped"/"segmentation fault"),
		sometimes causing R to crash. 
		Fixed.

	[6] rmh
		If the message 'increasing storage space and continuing'
		was issued, the algorithm restarted without updating its
		internal random number generator. This is not believed to
		have caused problems, because proposal points were 
		generated externally. 
		Fixed.