# $Id: CHANGES,v 1.12 2002/07/03 08:14:18 hothorn Exp $

	0.5-9 (03.07.2002)

	use .C(..., PACKAGE="maxstat")

	0.5-7 (22.05.2002)

	the cutpoints are in indices (which(!ties) - 1) -> Fixed

	0.5-6 (24.11.2002)

	iscores argument added (use integer valued scores or not)

	0.5-5 (11.04.2002)

	replace optim by optimize, passes R-1.5.0pre checks

	0.5-4 (06.03.2002)

	link preprint

	0.5-3 (14.12.2001)

	fixed bug in pLausen94: take tied predictors into account

	0.5-1 (11.12.2001)

	check the return value of require in zzz.R

	0.5-0 (10.12.2001)

	release for R-1.4.0

	0.4-6 (08.12.2001)

	cosmetrics, \dontrun{ postscript() } in DLBCL.Rd

	0.4-5 (04.12.2001)

	interface changes in mvtnorm, 
	Depends: exactRankTests(>= 0.5-8),mvtnorm(>= 0.4-1)


	0.4-4 (03.12.2001)

	survival interface now as in package survival
	pmaxstat returns a double, not a list
	man-pages updated

	0.4-3 (01.12.2001)

	clean interface (for survival problems too)
	no longer depends on survival

	0.4-2 (01.12.2001)

	maxstat.test generic passed R CMD check
	plot.maxtest and print.maxtest implemented

	0.4-1 (29.11.2001)

	maxstat.test is now generic. 
	To-Do:
	the interface is not clean for survival
	problems (formula for cens???, `Surv' isn't nice for that).

	take off `plot' and use `retT' to get the empirical process plotted
	(including p-values as in strucchange :-)

	for the examples: check, if `survival' is there and don't Depend on
	it

