2009-12-01  Martin Maechler  <maechler@stat.math.ethz.ch>

	* DESCRIPTION (Version): 0.9-9 - just for this fix:

	* R/l1looDesign.R (loo.design): s/z1.1/z1.2/ typo

2009-11-02  Martin Maechler  <maechler@stat.math.ethz.ch>

	* DESCRIPTION (Version): 0.9-8 - just for this fix:

	* man/cobsOld.Rd, cobs.Rd, ..: fix old syntax errors that newer versions of R
	(Rd_parse) no longer accept.

2007-10-09  Martin Maechler  <maechler@stat.math.ethz.ch>

	* man/globtemp.Rd: and others: eliminate "\non_function"
	and release 0.9-7 to CRAN

2006-08-03  Martin Maechler  <maechler@stat.math.ethz.ch>

	* DESCRIPTION (Version): 0.9-6 -- released to CRAN

	* src/drqssbc.f (drqssbc): replace the "^L" (not liked by gfortran 4.1)
	by outline minor mode and outline-regexp (in emacs).

2006-07-31  Martin Maechler  <maechler@stat.math.ethz.ch>

	* R/l1looDesign.R (l1.design): fixed ***LONG***standing bug:
	  for  (lambda = *, degree=1, constraint="none", pointwise = *),
	  l1.design() could not work.

	* tests/ex3.R: test the above case (and similar ones)

2006-07-07  Martin Maechler  <maechler@stat.math.ethz.ch>

	* R/cobs.R (cobs): explicit (instead of "hidden") default
		nknots = if(lambda == 0) 6 else 20; no warning any more
		when user choses nknots (or 'knots') differently.

	* DESCRIPTION (Version, Date): 0.9-6

	* R/cobs.R (cobs): ifl=3 from qbsks() must give error, not just warning.
	* R/qbsks.R (qbsks): mention 'nknots' from "knot selection" message.
		when some 'ifl' are not ok; improve determination of best,
		and of warning/error messages otherwise.
	* src/drqssbc.f: cosmetic:  white space  &  if(t1 < t & t < t2)


2006-04-04  Martin Maechler  <maechler@stat.math.ethz.ch>

	* DESCRIPTION (Version, Date): 0.9-5 for CRAN:
		trivial change: move R/README and R/.R-porting to toplevel

2003-11-07  Martin Maechler  <maechler@stat.math.ethz.ch>

	* DESCRIPTION (Version, Date): 0.9-4 for CRAN

	* R/cobsOld.R (cobsOld): eliminate multi-value return()
	* R/cobs.R (getdim):	<ditto>
	* R/drqssbc.R (drqssbc): <ditto>
	* R/qbsks.R (qbsks): 	<ditto>
	* R/cobsOld.R (cobsOld): give PACKAGE even in old code.

	* man/drqssbc.Rd: for "codoc": add same default values as in *.R

2002-08-28  Martin Maechler  <maechler@stat.math.ethz.ch>

	* src/dblas1.f (srotm1): declare sx(*) not sx(1)

2002-04-18  Martin Maechler  <maechler@stat.math.ethz.ch>

	* tests/README: new file; also renamed *.Rout.save to -N-save
	                (such that they are *not* compared).

	* DESCRIPTION (Version): 0.9-3  --> first version to CRAN

	* README: mention other READMEs, etc

2002-04-15  Martin Maechler  <maechler@stat.math.ethz.ch>

	* inst/scripts/README: new files: Three examples from Hg & Ng (1999)
		the installed scripts should become shorter than the ./tests/ ones

	* R/cobs.R (summary): slight improvement; R^2 only for median
	* R/cobs.R (mesg): eliminated (improves readability)

	* R/l1looDesign.R (l1.design, loo.design):
	* R/drqssbc.R (drqssbc):
	* R/qbsks.R (qbsks):      new files (splitting R/cobs.R)

2002-04-13  Martin Maechler  <maechler@stat.math.ethz.ch>

	* tests/small-ex.R: new file

	* R/cobs.R (predict.cobs): use crossprod(), colSums()
	(shat): make sure bound stays in [0,1]

2002-04-12  Martin Maechler  <maechler@stat.math.ethz.ch>

	* DESCRIPTION (Version): 0.9-2  -- since we *have* problem, see, e.g.:
	* src/README: added comment about roof example problem

	* tests/wind.R:  Added the three examples stricts,
	* tests/temp.R:  slightly improved to the tests directory
	* tests/roof.R:  ___ not yet using any *.Rout.save !! __

2002-04-01  Martin Maechler  <maechler@stat.math.ethz.ch>

	* R/splines.R (.splBasis): moved to this new file


2002-03-26  Martin Maechler  <maechler@stat.math.ethz.ch>

	* data/globtemp.R,
	* data/USArmyRoofs.tab,
	* data/DublinWind.tab: The 3 example data sets from the COBS paper

2002-02-27  Martin Maechler  <maechler@stat.math.ethz.ch>

	* DESCRIPTION (Version): 1.0-1  -- `pre-released' MM's FTP dir.

2002-02-20  Martin Maechler  <maechler@stat.math.ethz.ch>

	* src/splines.c: protoized (now ANSI headers!)

	* R/cobs.R (predict.cobs): allow out-of-knots prediction.
	  this only works in some cases!

	* R/cobs.R (loo.design) & al.: added  PACKAGE = "cobs"
	  (necessary when library(splines) is loaded with *same* routine names!)

2002-02-19  Martin Maechler  <maechler@stat.math.ethz.ch>

	* R/cobs.R (qbsks), (cobs) and (drqssbc): new argument `eps'
	  (with back-compatible default).

	* tests/ex1.R,
	* tests/ex1Old.R: added "lstart = 7872", to make both compatible

	* man/cobs-methods.Rd:
	* R/cobs.R (fitted.cobs), (residuals.cobs): new methods.

2002-02-18  Martin Maechler  <maechler@stat.math.ethz.ch>

	* R/cobs.R (summary.cobs): new method, print.cobs();
	predict.cobs() now `works' (NOT identically to cobsOld() !bug??)

2002-02-16  Martin Maechler  <maechler@stat.math.ethz.ch>

	* R/cobsOld.R (cobsOld): new file for cobsOld() function
	* R/cobs.R (cobs): newly return "cobs" object

	* src/monitor.c: new printing of diagnostics during Fortran

2002-02-15  Martin Maechler  <maechler@stat.math.ethz.ch>

	* src/README: new file showing call tree of Fortran code

	* tests/dunif01.R: new file; some exploration of COBS-builtin RNG
	* src/dunif01-tst.c: new: for the above

2002-02-14  Martin Maechler  <maechler@stat.math.ethz.ch>

	* R/cobs.R (cobs): equal, smaller,.. : simplified ->
	  instead of NULL, matrices w/ 0 rows  ==> change in other functions
	  (qbsks) and others: get rid of warn() but do the "warnings"

2002-02-13  Martin Maechler  <maechler@stat.math.ethz.ch>

	* man/cobs-internal.Rd: new file; now "R CMD check" is perfect

	* tests/ex1.R: new file; test basic example;

	* man/exHe.Rd: example fine

	* man/cobs.Rd: example improved

	* R/cobs.R (cobs): fix nknots to be smaller when unique(xo) is
	  (this currently leads to a Warning and a seg.fault

	  Added ``labels = FALSE'' to calls to cut() in order to get integers
