TODO (mostly to myself, some in German)
=====

12) tests/artif3.R : the Sig() is "silly": The covariance matrix has an
    ~~~~~~~~~~~~~~   Eigen decomposition where some of the eigen vectors
    are very ill-determined ==> platform dependent :
     ./tests/artif3.Rout-32b vs ./tests/artif3.Rout-64b
    ===> use another Sig() !

2)  jetzige Print methode print.simpcomp() schreibt sehr viel
    ==> Vorschlag, dies wird "summary()",
    und ich schreibe Methode mit krzerem Output.

5) Etliche der Hilfe Seiten fr die "Low Level" Funktionen,
   sind noch praktisch leer, haben auch keine Beispiele:

   File			     Fehlende Dok. | nur fehlende "Examples"
   ----			     -------------   -----------------
   ./sca/man/agglomblock.Rd			x
   ./sca/man/allcrit.Rd				x
   ./sca/man/firstpcres.Rd			x
   ./sca/man/maxmatrix.Rd			x
   ./sca/man/nextdiff.Rd			x
   ./sca/man/quickcrit.Rd			x
   ./sca/man/shrinkdiff.Rd			x
   ./sca/man/sortmatrix.Rd			x


6) quickcrit(), firstpcres(), ...
   FIXME:  Lots of speed to be gained by smarter    code / formulae!
    -- done partly: using crossprod(); solve() with RHS

8) The paper (?) and the "redrawMatrix" output use the word `stage',
   whereas the S code uses phase ---> rename phase to stage !

9) Now (2003-07-09) that I've looked at the S-plus version, I've realized
   how nicely chosen sca()'s colors are there.
   The R version definitely needs better coloring!

10) sca() & redrawMatrix() : When p (= ncol(.)) is larger,
    the R version should  use a smaller cex or tell the user to first
    enlarge the graphics window {or do both!} instead of failing.

11) R/aaa.R  is not used in R, but rather for the S-plus version.
    -------  use a different name (ending) such that R does not even source
    it? --> this would get rid of the (wrong) extraneous warnings with
    R 1.8's "R CMD check".
      --- partly done (AFAICS even all: no warnings in R-devel-2004-08-06)
