[Rd] Matrix package breaks as.matrix method
Robert McGehee
rmcgehee at walleyetrading.net
Mon Jan 30 15:38:02 CET 2017
Georgi,
Brilliant, thank you very much for the helpful reply and explanation! I added 'importFrom("Matrix","as.matrix")' to my NAMESPACE and all worked fine! As my 'as.matrix' method is used entirely internally to the 'testmat' function (and not "used outside the package"), I don't think I actually need to export it. In my case, testmat is defined inside the package, and not in the global workspace (nothing is defined in the global workspace in my example).
To your point that I should just "import the 'Matrix'", I thought adding Matrix to the Depends field in the DESCRIPTION would do that for me, but apparently I need to study the WRE manual more thoroughly, as it clearly does not.
It's also worth pointing out that R CMD check did not actually warn that I should have imported as.matrix, at least in the case where I have 'Depends: Matrix' in DESCRIPTION and only 'exportPattern("^[[:alpha:]]+")' in the NAMESPACE file.
Last, for anyone who took issue, apologies for implying a bug where none exists, or that there was something wrong with Matrix or S4.
Thanks again,
Robert
-----Original Message-----
From: Georgi Boshnakov [mailto:georgi.boshnakov at manchester.ac.uk]
Sent: Sunday, January 29, 2017 1:40 PM
To: r-devel at r-project.org
Cc: Robert McGehee <rmcgehee at walleyetrading.net>
Subject: RE: R-devel Digest, Vol 167, Issue 25
Hi,
Short answer: import 'as.matrix' and export your method(s) for it. From WRE:
"All S4 classes to be used outside the package need to be listed in an exportClasses directive. Alternatively, they can be specified using exportClassPattern.(46) in the same style as for exportPattern. To export methods for generics from other packages an exportMethods directive can be used."
Details: the precise details depend on what exactly is in your NAMESPACE file.
The curious difference you observe is due to the fact that as.matrix is defined in 'base' as S3 generic.
When you set an S4 method for it you effectively create your own as.matrix S4 generic and your settings in NAMESPACE probably export everything you "own".
When you depend on package 'Matrix', 'as.matrix' is made S4 generic by it and you are defining a method for An imported function, so you need to use exportMethods (see the above excerpt), otherwise your method is not exported. (Indeed, in your example the arror is from array() in the default method).
Note also that it matters where you define the test function. I am pretty sure that you defined it in the global workspace (not in the package) to get the error.
It is very useful practice to run 'R CMD check' (or its devtools equivalent). In this case it would have warned you to import as.matrix. Also, to make your life easier during initial development, import the whole 'Matrix' and when things work think about cleaning up and importing only stuff that you need.
Best regards,
Georgi Boshnakov
------------------------------
Message: 15
Date: Fri, 27 Jan 2017 22:41:10 +0000
From: Robert McGehee <rmcgehee at walleyetrading.net>
To: "r-devel at r-project.org" <r-devel at r-project.org>
Subject: [Rd] Matrix package breaks as.matrix method
Message-ID:
<30D28A63376088428E8318DD67FD407F705B2E at ny-mailstore1.walleyetrading.net>
Content-Type: text/plain; charset="us-ascii"
Hi,
The Matrix package and the as.matrix method do not seem to be compatible inside of a package.
Here's an example. I've created a simple package "mat" that defines an eponymous class and as.matrix method on that class. All is well, unless that package has the Matrix package in its Depends or Imports (and imports, e.g. the "Diagonal" function). Then my as.matrix method stops working, even if I'm not using any part of the Matrix package.
Here's an example on R 3.3.2:
First, create an empty package "mat" (e.g. with package.skeleton) with one file in mat/R/mat.R with the following contents:
setClass("mat", representation(H="matrix")) mat <- function(H) new("mat", H=H) setMethod("as.matrix", signature("mat"), function(x, ...) crossprod(x at H)) testmat <- function() {
H <- matrix(1:3, 1, 3)
M <- mat(H)
as.matrix(M)
}
Then install the mat package :
> require(mat)
> testmat()
[,1] [,2] [,3]
[1,] 1 2 3
[2,] 2 4 6
[3,] 3 6 9
All works fine!
Now add "Depends: Matrix" into the package's DESCRIPTION file (alternatively add "Imports: Matrix" and 'importFrom("Matrix","Diagonal")' in the NAMESPACE).
Try again:
> require(mat)
> testmat()
Error in as.vector(data) :
no method for coercing this S4 class to a vector
Bug? If not, can anyone provide a work around? In my case, I'd like to mix matrix and Matrix functions in my package, but am obviously having difficulty.
I've come across a somewhat similar report on stackoverflow http://stackoverflow.com/questions/13812373/overloading-operator-in-r-s4-classes-and-matrix-package regarding defining the "+" operator with the Matrix package, but I don't think the solution or the problem quite applies.
Thanks in advance, Robert
> R.version
_
platform x86_64-pc-linux-gnu
arch x86_64
os linux-gnu
system x86_64, linux-gnu
status
major 3
minor 3.2
year 2016
month 10
day 31
svn rev 71607
language R
version.string R version 3.3.2 (2016-10-31)
nickname Sincere Pumpkin Patch
------------------------------
Message: 16
Date: Fri, 27 Jan 2017 18:52:33 -0800
From: Henrik Bengtsson <henrik.bengtsson at gmail.com>
To: R-devel <r-devel at r-project.org>
Subject: [Rd] Subject: Milestone: 10000 packages on CRAN
Message-ID:
<CAFDcVCRC6pp3Humv7o48AerZ10Ox0bnxuo3COydzaGfCY17DUA at mail.gmail.com>
Content-Type: text/plain; charset=UTF-8
Continuing the tradition to post millennia milestones on CRAN:
So, it happened. Today (January 27, 2017 PCT) CRAN reached 10,000 packages [1].
Needless to say, the rate with which new packages are added to CRAN keeps increasing and so does the number of contributors (maintainers).
Somewhere out there, there are ~3 persons who are about to submit their first packages to CRAN today and ~3 persons who will submit another package of theirs. And by the amazing work of the CRAN team, these packages are inspected and quality controlled before going live
- which often happens within a day or so.
As usual and it can't be said too many times: A big thank you to the CRAN team, to the R core, to all package developers, to our friendly community, to everyone out there helping others, and to various online services that simplify package development. We can all give back by carefully reporting bugs to the maintainers, properly citing packages we use in publications (see citation("pkg")), and help newcomers to use R.
Milestones:
2017-01-27 10000 pkgs (+6.3/day over 158 days) 5845 mnts (+3.5/day)
2016-08-22 9000 pkgs (+5.7/day over 175 days) 5289 mnts (+5.8/day)
2016-02-29 8000 pkgs (+5.0/day over 201 days) 4279 mnts (+0.7/day)
2015-08-12 7000 pkgs (+3.4/day over 287 days) 4130 mnts (+2.4/day)
2014-10-29 6000 pkgs (+3.0/day over 335 days) 3444 mnts (+1.6/day)
2013-11-08 5000 pkgs (+2.7/day over 442 days) 2900 mnts (+1.2/day)
2012-08-23 4000 pkgs (+2.1/day over 469 days) 2350 mnts
2011-05-12 3000 pkgs (+1.7/day over 585 days)
2009-10-04 2000 pkgs (+1.1/day over 906 days)
2007-04-12 1000 pkgs
2004-10-01 500 pkgs
2003-04-01 250 pkgs
2002-09-17 68 pkgs
1997-04-23 12 pkgs
These data are for CRAN only [1-13]. There are many more packages elsewhere, e.g. Bioconductor, GitHub, R-Forge etc.
[1] http://cran.r-project.org/web/packages/
[2] https://en.wikipedia.org/wiki/R_(programming_language)#Milestones
[3] http://www.r-pkg.org/
[4] Private data
[5] https://stat.ethz.ch/pipermail/r-devel/2007-April/045359.html
[6] https://stat.ethz.ch/pipermail/r-devel/2009-October/055049.html
[7] https://stat.ethz.ch/pipermail/r-devel/2011-May/061002.html
[8] https://stat.ethz.ch/pipermail/r-devel/2012-August/064675.html
[9] https://stat.ethz.ch/pipermail/r-devel/2013-November/067935.html
[10] https://stat.ethz.ch/pipermail/r-devel/2014-October/069997.html
[11] https://stat.ethz.ch/pipermail/r-package-devel/2015q3/000393.html
[12] https://stat.ethz.ch/pipermail/r-devel/2016-February/072388.html
[13] https://stat.ethz.ch/pipermail/r-devel/2016-August/073011.html
All the best,
Henrik
(just a user)
------------------------------
Subject: Digest Footer
_______________________________________________
R-devel at r-project.org mailing list DIGESTED https://stat.ethz.ch/mailman/listinfo/r-devel
------------------------------
End of R-devel Digest, Vol 167, Issue 25
More information about the R-devel
mailing list