TODO

-*- org -*-

Priority 1-2 (urgent - important)

REML <-> ML ‘Std.Error’s for fixed effecs in summary(n?lme()) are the same but should not,

Both S+ nlme [note from Siem Heisterkamp] and e.g. lme4 do differ.

svn ci -m’patch from Siem Heisterkamp [‘apVar’ for “fixed sigma”] + changes to make intervals(.) working’

svn-diffB R/gls.R R/gnls.R R/lme.R R/newFunc.R R/nlme.R man/glsControl.Rd man/gnlsControl.Rd man/gnlsObject.Rd man/intervals.lmList.Rd man/intervals.lme.Rd man/lmeControl.Rd man/lmeObject.Rd man/nlmeControl.Rd man/nlmeObject.Rd tests/sigma-fixed-etc.R

svn ci -m’fix qqnorm.lme() bug - undetected because of “wrong” \dontrun{}’

svn-diffB R/lme.R man/qqnorm.lme.Rd

Priority 3-4 (project for …)

stats::summary.lm , stats::print.summary.lm by default do NOT

compute and print the correlation matrix.

stats:::summary.lm (object, correlation = FALSE, symbolic.cor = FALSE, …)

stats:::print.summary.lm (x, digits = max(3L, getOption(“digits”) - 3L),

symbolic.cor = x$symbolic.cor, signif.stars = getOption(“show.signif.stars”), …)

‘nlme’ should at least allow to not compute and print

so people would use summary(.) much more often probably ! MM thinks the default should even change not to print the correlation

Bug 16703 – corFactor/corMatrix use O(n^2) storage - sparseMatrix would help

https://bugs.r-project.org/show_bug.cgi?id=16703

Bug 16806 - nlme getVarCov() fails for corSpatial fits for individuals with a single observation only

https://bugs.r-project.org/show_bug.cgi?id=16806

methods extractAIC.lme & extractAIC.gls should be here rather than in MASS,

because extractAIC() and step() are in stats

Priority 5 (nice to have)