[R] R-3.4.0 and survival_2.41-3 ..

[Göran Broström]

On 2017-04-25 10:34, Martin Maechler wrote:
>>>>>> Göran Broström <goran.brostrom at umu.se>
>>>>>>     on Tue, 25 Apr 2017 10:22:48 +0200 writes:
>
>     > I installed R-3.4.0 and got problems with the survival package, for instance
>     > ------------
>     >> library(survival)
>     >> mort <- data.frame(exit = 1:4, event = rep(1, 4), x = c(0, 1, 0, 1))
>     >> fit <- coxph(Surv(exit, event) ~ x, data = mort)
>     > Error in fitter(X, Y, strats, offset, init, control, weights = weights,  :
>     > object 'Ccoxmart' not found
>     > -------------
>
>     > No problems with R-3.3.3 and the same (latest) survival version, which
>     > makes me think that something is going on in my R installation rather
>     > than in the survival package.
>
>     > Thanks for any hint,
>
>     > Göran
>
>     >> sessionInfo()
>     > R version 3.4.0 (2017-04-21)
>     > Platform: x86_64-pc-linux-gnu (64-bit)
>     > Running under: Ubuntu 16.04.2 LTS
>
>     > Matrix products: default
>     > BLAS: /usr/lib/openblas-base/libblas.so.3
>     > LAPACK: /usr/lib/libopenblasp-r0.2.18.so
>
>     > locale:
>     > [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
>     > [3] LC_TIME=sv_SE.UTF-8        LC_COLLATE=en_US.UTF-8
>     > [5] LC_MONETARY=sv_SE.UTF-8    LC_MESSAGES=en_US.UTF-8
>     > [7] LC_PAPER=sv_SE.UTF-8       LC_NAME=C
>     > [9] LC_ADDRESS=C               LC_TELEPHONE=C
>     > [11] LC_MEASUREMENT=sv_SE.UTF-8 LC_IDENTIFICATION=C
>
>     > attached base packages:
>     > [1] stats     graphics  grDevices utils     datasets  methods   base
>
>     > other attached packages:
>     > [1] survival_2.41-3
>
>     > loaded via a namespace (and not attached):
>     > [1] compiler_3.4.0  Matrix_1.2-8    splines_3.4.0   grid_3.4.0
>     > [5] lattice_0.20-35
>
> I'm 99.5% sure that you are using more than just the default
> library of package (a very good thing - we have been doing the
> same for years).
>
> We have in NEWS for R 3.4.0
>
>   > PACKAGE INSTALLATION:
>
>   >   [...........]
>
>   >   [...........]
>
>   >   • Packages which register native routines for .C or .Fortran need
>   >     to be re-installed for this version (unless installed with
>   >     R-devel SVN revision r72375 or later).
>
> and Prof Brian Ripley did announce that nicely and early on R-devel.
> ==> https://hypatia.math.ethz.ch/pipermail/r-devel/2017-March/073940.html
>
> ==> You have to re-install quite a few packages for R 3.4.0,
>      __if__ they use .C() or .Fortran()
>
> When we've e-talked about the issue within R-core,
> Uwe Ligges noted we should really ask everyone to run
>
>    	  update.packages(checkBuilt=TRUE)

A small nuisance with this is that I end up with two versions of the 
survival package (and other recommended packages), since I cannot write 
to /usr/lib/R/library. However, this is obviously a problem suitable to 
present on R-SIG-Debian. See you there.

Göran

>
> after an update to a new major (meaning "R-x.y.0") release of R
> and **not** re-use packages {inside R-x.y.z}
> that were installed with R-x.(y-1).z'  ..
>
> and of course Uwe is right:
> We should ask others to do it _and_ do it ourselves.
>
> Anyway it _is_ considerably more important for the 3.4.0
> release.
>
> Martin Maechler
> ETH Zurich (and R Core team)
>
>