[R] lme4 + R 2.11.0 + mac unavailable

Martin Maechler maechler at stat.math.ethz.ch
Wed May 19 12:36:18 CEST 2010

Just a small correction to what I've said ... see below

>>>>> "MM" == Martin Maechler <maechler at stat.math.ethz.ch>
>>>>>     on Tue, 18 May 2010 10:00:20 +0200 writes:

>>>>> David Winsemius <dwinsemius at comcast.net>
>>>>>     on Mon, 17 May 2010 17:44:00 -0400 writes:

    >> On May 17, 2010, at 5:01 PM, Adam November wrote:

    >>> I believe I'm working with the newest version of R
    >>> (2.11.0) and I've tried a few of the most recent
    >>> versions... No luck yet.

    >> Luck has nothing to do with it. What part of ERROR in the
    >> CRAN package check for the current version of lme4 with
    >> the Mac R 2.11.0 don't you understand? Packages that
    >> generate errors are not made available as binaries.

    >> I was giving you advice about how to deal with the stated
    >> difficulty with installing lme4 from source using the
    >> older versions of R. Only if you want to revert to 2.10.1
    >> and install a prior version for lme4 would my advice be
    >> helpful.

    >> -- 
    >> David.

    MM> Doug Bates and I had recently been contacted by another Mac user
    MM> about this.
    MM> Doug advised him to install  lme4  from the sources and
    MM> he was immediately successful.

    MM> In the light of this, I'm a bit puzzled about the harsh words
    MM> David is using with respect to the "cohabitation" of R 2.11.0
    MM> and lme4.

    MM> Simultaneously, we have been in contact with Simon Urbanek
    MM> trying to investigate why the 'R CMD check' tests fail on OSX
    MM> (http://www.r-project.org/nosvn/R.check/r-release-macosx-ix86/lme4-00check.html)
    MM> A conclusion/solution has not been found yet,
    MM> but as I said above, the advice to install from the source

    MM> install.packages("lme4", type = "source")

The following is not true.  There is basically no caveat here.

Yesterday I had managed to *not* get lme4 to install because I
had too many directories called 'Matrix' "lying around"  which
confused package installation for lme4 when it was looking for
the Matrix header files.
This should never happen to "normal" people ...

    MM> One possible caveat -- as lme4 uses the C API of Matrix (including
    MM> the C header ("include") files it exports) --
    MM> is that for the above way of installation,
    MM> you need to install into the (or "a") library where Matrix is
    MM> installed.
    MM> For those who only use one library, this will not be
    MM> a concern.  Others could use the fact that MacOSX is a kind of Unix
    MM> and you can simply "symlink" Matrix (from the R standard
    MM> library) into the library into which you want to install lme4
    MM> {  cd <mylibrary>
    MM> ln -s `R RHOME`/library/Matrix . 
    MM> }

    MM> Martin Maechler, ETH Zurich

    >>> Thanks, -Adam
    >>> On Mon, May 17, 2010 at 1:51 PM, David Winsemius
    >>> <dwinsemius at comcast.net > wrote:
    >>>> On May 17, 2010, at 4:44 PM, Adam November wrote:
    >>>>> Hi All, Just thought I'd bring attention to the fact
    >>>>> that lme4 is failing cran checks on the mac platform,
    >>>>> and I can't seem to install it from source on 10.5 or
    >>>>> 10.6, either ("ld: library not found for -lgfortran ")
    >>>>> .  Any help getting this working?  Thanks!
    >>>> You would not expect the current package to install
    >>>> properly with earlier versions of r. Find the
    >>>> appropriate version at the archive:
    >>>> http://cran.r-project.org/src/contrib/Archive/lme4/
    >>>> --
    >>>> David Winsemius, MD West Hartford, CT

    MM> ______________________________________________
    MM> R-help at r-project.org mailing list
    MM> https://stat.ethz.ch/mailman/listinfo/r-help
    MM> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
    MM> and provide commented, minimal, self-contained, reproducible code.

More information about the R-help mailing list