[R-pkg-devel] Migrating a package name

[Lenth, Russell V]

Hello,

I have a package 'lsmeans' (for "least-squares means") for follow-up analyses of models; and it has become increasingly clear to me that its name is archaic and misleading. It works for GLMs, ordinal and count regression, Bayesian models fitted using MCMC, and all sorts of other models that may or may not have been fitted by least squares. A better term for what the package does is "estimated marginal means" -- a term has become widely accepted. 

In my own recent development version of 'lsmeans', I have changed all kinds of internal stuff, and the primary user functions are now emmeans() and such. I also have written wrappers so that an existing user can still run the same analyses they did before with the same function calls. For example, the lsmeans() function runs emmeans() and changes the column labels in the results. The wrappers actually work: all the old example code and vignette examples run and produce the same results as before. But 'lsmeans::emmeans()' just doesn't look good. I'd like to rename the package 'emmeans'.

Here's what I propose to do:

  -- After thorough checking, submit a new 'emmeans' package to CRAN that has the same (a bit more) functionality as the current 'lsmeans'.

  -- Shortly thereafter, submit a revised 'lsmeans' package that consists just of the wrappers and minimal documentation. It will depend on 'emmeans'.

  -- Perhaps eventually remove 'lsmeans' (but it will be a very small package anyway)

This ensures continuity for current 'lsmeans' users, allows for migrating to the new name, and avoids having two packages on CRAN that do the same thing, except for the brief transition period.

Are there any additional issues I need to consider, and is this OK per CRAN policies?

Russ

Russell V. Lenth  -  Professor Emeritus
Department of Statistics and Actuarial Science   
The University of Iowa  -  Iowa City, IA 52242  USA   
Voice (319)335-0712 (Dept. office)  -  FAX (319)335-3017