fPortfolio calls internals of quadprog which have changed
Yohan Chalabi
chalabi at phys.ethz.ch
Sat Apr 17 13:59:47 CEST 2010
>>>> "PBR" == Prof Brian Ripley <ripley at stats.ox.ac.uk>
>>>> on Sat, 10 Apr 2010 08:06:38 +0100 (BST)
PBR> You will see failures for BLCOP and fPortfolio on the more
PBR> recent
PBR> entries on the CRAN Package checks at
PBR> http://cran.r-project.org/web/checks/check_summary.html
PBR>
PBR> E.g.
PBR>
PBR>
http://www.r-project.org/nosvn/R.check/r-prerel-linux-x86_64-gcc-fedora/BLCOP-00check.html
PBR> has PBR>
PBR> # File
PBR> /data/gannet/tmp/ripley/R/packages/tests-devel/BLCOP.Rcheck/inst/doc/BLCOP.Rnw
PBR> :
PBR> chunk 7
PBR> Error in .Fortran("qpgen2", as.double(Dmat), dvec =
PBR> as.double(dvec),
PBR> as.integer(n), :
PBR> Incorrect number of arguments (16), expecting 17 for qpgen2
PBR>
PBR> The issue is that fPortfolio is calling the .Fortran entry
PBR> point in
PBR> quadprog. This really is not a good idea, so can you
PBR>
PBR> - urgently put out a revision to fPortfolio which addresses
PBR> this and
PBR> depends on quadprog >= 1.5-2 (or whatever is needed)
PBR>
PBR> - see if it is possible to work with the API functions in
PBR> quadprog or
PBR> if not negotiate with Berwin to add an R function wrapper
PBR> that does
PBR> what you need.
PBR>
PBR> Brian Ripley
Thanks for your comments.
Please find a new version of fPortfolio on CRAN.
Regards,
Yohan
--
PhD candidate
Swiss Federal Institute of Technology
Zurich
www.ethz.ch
More information about the Rmetrics-core
mailing list