2001-10-08  Douglas Bates  <bates@stat.wisc.edu>

	* R/lmList.R: Fixed definition of logLik.lmList to work with R
	(but it is still rather ugly).

	* R/varFunc.R: Changed definition of logLik.varComb to return a
	logLik object.

2001-10-05  Douglas Bates  <bates@stat.wisc.edu>

	* R/lmList.R: Fixed selection of id[subscripts] in the panel
	function of pairs.lmList (and several other functions) so it does
	not produce a warning from evaluating is.na(NULL).

2001-09-17  Douglas Bates  <bates@stat.wisc.edu>

	* R/nlme.R: Incorporated a change from
	Setzer.Woodrow@epamail.epa.gov for using nested grouping factors
	in nlme.

2001-06-20  Douglas Bates  <bates@stat.wisc.edu>

	* R/varFunc.R: Changes in calling sequences and in documentation
	so the package passes codoc() checks.  These affect many files in
	both the R and the man directory.  No changes of substance.

2001-05-29  Martin Maechler <maechler@stat.math.ethz.ch>

	* R/zzMethods.R: Use .Alias() 
	
	* R/newGenerics.R:  Generic AIC()
	* R/newMethods.R:   logLik.lm()   moved to "base R"
	* man/AIC.Rd,   
	* man/AIC.logLik.Rd: AIC.logLik() and AIC.lm() as well
	* man/logLik.lm.Rd: 

2001-03-28  Douglas Bates  <bates@stat.wisc.edu>

	* R/simulate.R: Modified simulate.lme for default pdMat class of
	pdLogChol.

2001-02-24  Douglas Bates  <bates@stat.wisc.edu>

	* R/newFunc.R: Removed the old fake xyplot function.  The new work
	by Murrell and Sarkar will make this superfluous.

	* man/MEestimate.Rd: Added manual pages for MEEM, MEdecomp,
	MEestimate.

	* R/newMethods.R: Revised assignments in plotting methods to be
	compatible with the Murrell/Sarkar preliminary trellis package.

	* src/nlmefit.c (internal_decomp): Corrected an error reported by
	Kathryn Chaloner where Srows >= ZXrows.

2001-01-10  Douglas Bates  <bates@stat.wisc.edu>

	* DESCRIPTION (Maintainer): Added Maintainer entry to
	DESCRIPTION. 

2000-12-09  Douglas Bates  <bates@stat.wisc.edu>

	* R/lme.R: (Variogram.lme) Coerced n.pairs to numeric and
	corrected calls to dist.

2000-12-08  Douglas Bates  <bates@stat.wisc.edu>

	* R/lme.R: Changed all calls to dist in lme.R, gls.R and
	corStruct.R to use the argument "method" rather than "metric".

	* R/pdMat.R: Replaced calls to .Fortran("dbksl", ...) with calls
	to the R function solve() for R-1.2.0 where the S_compat.c file
	has been removed.

	* inst/scripts/ch01.R: Added the chapter scripts from
	"Mixed-Effects Models in S and S-PLUS" to the inst directory.

	* man/formula.nlme.Rd: Modified the examples in this and many
	other .Rd files to use the Loblolly data and the SSasymp
	self-starting function.  There were many occasions where the
	Soybean data was being fit to a default formulation of an nlme
	model just to get things like a formula.  This fit of the Soybean
	data is difficult and time consuming and not very illuminating.

2000-11-30  Douglas Bates  <bates@stat.wisc.edu>

	* src/matrix.c: Removed dependency on S_compat.h by replacing
	calls to dqrdca with the equivalent call to dqrdc2.
	
2000-07-24  Saikat DebRoy <saikat@stat.wisc.edu>

	* Changed various uses of do.call to eval.  The reason for doing
	this is to allow calls to the function calling eval from within
	another function.  It's the old scoping problem again.

2000-07-07  Saikat DebRoy <saikat@stat.wisc.edu>

	* src/nlme.c, src/nlmefit.c: Added mixed_grad in calls to optif9.

	* src/nlmefit.c : Added handling of pdLogChol objects (pdClass =
	4) in various functions.

	* src/nlmefit.c (mixed_grad): New function giving analytic
	gradient for mixed-effects log-likelihood.

	* R/nlme.R (nlme.formula): Added updating of the pdFactor in the
	PNLS step. 

	* R/reStruct.R: Made pdLogChol the default representation for
	reStruct objects.

	* src/pdMat.c (logChol_pd): new function.

	* R/pdMat.R: Added back the pdLogChol class and associated
	methods.
	

2000-07-03  Douglas Bates  <bates@stat.wisc.edu>

	* man/corFactor.corStruct.Rd: Added an alias for
	corFactor.corNatural

	* src/nlmefit.c (internal_loglik): Changed the return value on
	error from R_NegInf to -DOUBLE_XMAX so the optif[09] optimizers
	will work properly.

	* src/corStruct.c (corStruct_factList): Provided a correct call to
	the R version of F77_CALL(chol) here and everywhere else in this
	source file.

	* R/zzz.R: Removed the deprecated call to provide

2000-03-17  Douglas Bates  <bates@stat.wisc.edu>

	* added many data sets

	* incorporated patches from Brian Ripley to use optim for the
	optimization rather than nlm
	
	* src/nlmefit.c (internal_loglik): Returned R_NegInf instead of
	generating a error on a singular precision matrix.

	* DESCRIPTION (Depends): Changed Depends to R version 1.0.1
	because of problem with assigning components of lists.

1999-12-23  Douglas Bates  <bates@stat.wisc.edu>

	* TITLE: Changed the title to nlme

