[RsR] robustbase, names
claudio
c|@ud|o @end|ng |rom un|ve@|t
Thu Feb 2 11:19:08 CET 2006
On Thu, 2 Feb 2006, Peter Filzmoser wrote:
> I understand that moving functions to robustbase is only
> meant for "basic" functions, and not for more specialized
> methods. Eg. I don't think it would make much sense to
> merge the package mvoutlier into robustbase.
> In that case the pragmatic proposal of Martin is fine.
I agree with the pragmatic proposal of Martin. Perhaps, there could be a
warning message when the old package is load (or in robustbase or in
both) that say something along the lines:
the function FUNCTION is moving from package PACKAGE to package robustbase
in the next version, please use the function in the robustbase package for
your code.
Followed perhaps by a list of changed in the function call like
equivalence between arguments
> Concerning names: Both variants are attractive:
> 1) With "rob"... we can easiliy see which functions exist
> on robustness. However, this might be more interesting
> for "robustnics".
> 2) With ..."rob" it is easier for the user to guess the
> "robustified" version of what he/she is using all the
> time.
> Since robustnics are well informed about the progress
> of "Robustness and R" ;-) I would prefer the second variant.
I will prefere the second variant too, also because, you could have
covrobfoo (for a robust method foo in robust covariance estimate)
and
lmortho (for orthogonal regression in linear model)
and
lmorthorob (for robust method on orthogonal regression in linear model)
And as Peter said: Yes and - sure, I will also come to useR! and send an
abstract.
Claudio
More information about the R-SIG-Robust
mailing list