[R-sig-ME] Development version of lme4 has been released to CRAN

Martin Maechler maechler at stat.math.ethz.ch
Sat Jun 28 14:35:39 CEST 2008

>>>>> "MM" == Martin Maechler <maechler at stat.math.ethz.ch>
>>>>>     on Sat, 28 Jun 2008 14:07:25 +0200 writes:

>>>>> "KB" == Ken Beath <kjbeath at kagi.com>
>>>>>     on Sat, 28 Jun 2008 12:16:52 +1000 writes:

    KB> On 24/06/2008, at 5:47 AM, Douglas Bates wrote:

    >>> Early Sunday morning the development version of the lme4 package was
    >>> released to CRAN.  There has already been one update version released
    >>> and will be another later today.  The current CRAN version of the
    >>> source and the Windows binary package is 0.999375-17.
    >>> I expect that as users begin to install this package and try to run
    >>> old code they will encounter issues.  I will be traveling and have
    >>> limited email access from June 26 until July 2 and no email access
    >>> from July 3 until July 8.
    >>> I would appreciate it is those who are so helpful in responding to
    >>> questions about lme4 on this list and on the R-help list could be
    >>> extra vigilant during that period.  Some of the changes have been
    >>> documented in the NEWS file.

    KB> Lack of a MacOS version seems to be first that MEMSS should be depends  
    KB> rather than suggests, 

    MM> no, not at all:
    >> From "Writing R Extensions":

    >>> The optional `Suggests' field uses the same syntax as `Depends' and
    >>> lists packages that are not necessarily needed.  This includes packages
    >>> used only in examples or vignettes (*note Writing package vignettes::),

    MM> and it is true that you need the 'Suggests' packages when
    MM> *building* or *checking* the package itself.

    KB> then the Implementation.rnw vignette succeeds.  

    MM> (yes, see above)

    KB> Build then works for 0.999375-20 but check gives the following error

    >>> e3 <- expand(nm3)
    >>> stopifnot(identical(sapply(e3, class),
    KB> +                     c(sigma = "numeric", P = "pMatrix",
    KB> +                       T = "dtCMatrix", S = "ddiMatrix"))
    KB> +           , allEQ(e3$sigma, c(sigmaML = 0.76921295))
    KB> +           , all(e3$P at perm == outer(12*(0:2), 1:12, "+"))
    KB> +           , identical(as(e3$T, "diagonalMatrix"), Diagonal(3*12))
    KB> +           , allEQ(e3$S at x, rep(c(0, 0.620071, 0.163092), each=12))
    KB> +           )
    KB> Error: identical(as(e3$T, "diagonalMatrix"), Diagonal(3 * 12)) is  
    KB> not TRUE
    KB> Execution halted

    MM> This looks like you are not using the current version of
    MM> 'Matrix',  0.999375-10
    MM> which *is* required by  lme *-20, so I think you did not get
    MM> correct versions of one of the two packages.

Oops!  I had changed the Depends to   'Matrix (>= 0.999375-10)'
but unfortunately this was *not* committed before lme *-20 was
built.  In other words, the above  *is* is wrong.

What remains true is that
update.packages() will solve that one problem.

Being sorry for the added confusion,

      KB> MacOS X 10.5.3 Intel Core 2 Duo with R 2.7.1
      KB> Ken

More information about the R-sig-mixed-models mailing list