Prof Brian Ripley
2025-Sep-24 06:59 UTC
[Rd] Reviving IndependenceTests and LAPACK WARNINGs
From the NEWS for R 4.5.0:
? It is intended that this will be the last update to BLAS and
LAPACK in the R sources. Those building R from source are
encouraged to use external BLAS and LAPACK and this will be
required in future.
Your package could be using an external BLAS/LAPACK, or include the
routines you want in the package sources (if the licences allow).
On 20/09/2025 01:34, Pierre Lafaye de Micheaux via R-devel
wrote:> Dear R developpers,
> Two months ago, I sent an email to the CRAN team asking if they would
consider adding more LAPACK routines to the R's source tree. I have not
received a response from them, which led me to assume that maybe I used the
wrong communication channel. (I apologise if this is the case.)
> My goal is to revive my "IndependenceTests" package. It was
archived in 2021 (because I had not been able to solve various issues on time).
> I have now been able to resolve most of the issues, though I still WARNINGs
such as:
> Warning: GNU Extension: Nonstandard type declaration COMPLEX*16 at (1)
> It is my firm belief that I cannot solve these issues without modifying
several (old) Lapack Fortran source files.
> Now, in Section 6.7 of the "Writing R Extensions" manual, I read
that the current set of LAPACK subroutines included in R "might be
supplemented".
That is about those declared in Lapack.h, not about the routines
included in the R sources.
> Also, in the
https://github.com/SurajGupta/r-source/blob/master/src/modules/lapack/README
file,
That is not the official R sources and appears to be several years
out-of-date.
> I see that some Lapack files were added over the years to the R source
tree. But not all the ones I need, unfortunately.
> The file I would like to be added to the R source tree is the following:
> zhpevx.f
> which in turn needs those below that are not in the R Lapack:
> lsame.f
> xerbla.f
Those two are included -- lsame.f as part of the BLAS, xerbla is
implemented via a C version.
> zhptrd.f
> zlanhp.f
> zstein.f
> zupgtr.f
> zupmtr.f
> Would you kindly let me know if:
> 1.
> A formal request to add these LAPACK routines to the R's source tree is
something that can be considered?
See above.
> 2.
> Alternatively, would a submission with these few remaining warnings still
be acceptable?
It is not clear what those are. No CRAN package is showing COMPLEX*16
warnings in its checks although two do in their installation logs.
> I fully understand if this falls outside the scope of R Core Team or
CRAN's intended support, but any guidance would be greatly appreciated.
> Thank you in advance for your time and consideration.
> Kind regards,
> Associate Professor Pierre Lafaye De Micheaux
> UNSW Sydney
(Copyright message removed as used without permission. And the details
it linked to had expired.)
> [[alternative HTML version deleted]]
As the posting guide says, do not send HTML.
--
Brian D. Ripley, ripleybd at icloud.com
Emeritus Professor of Applied Statistics, University of Oxford