[R] Help with installing a .tar.gz package on windows

Duncan Murdoch murdoch.duncan at gmail.com
Fri Jul 12 17:38:43 CEST 2013

On 13-07-12 3:26 PM, Michael Dewey wrote:
> At 23:49 11/07/2013, Lucy Leigh wrote:
>> Hi everyone,
>> Thanks to everyone for all the advice. I should have been clearer in
>> my first email, the version of 'PReMiuM' I have is not
>> the one available on CRAN at the moment, but a version I was sent by
>> one of the authors, Silvia Liveriani, with a small bug fixed.
>> I don't know if this is the same as the next version to be released,
>> or whether the next will be different again. I asked about
>> compiling it myself because I didn't want to bother the authors
>> again if doing so was going to be a simple task I could
>> do myself, which I now realise is not the case.
>> I think the best option seems to me to be to contact the authors
>> again and request permission to send the version to
>> winbuilder to be compiled for me.
> Lucy
> The advantage of following Bill's instructions is that you will then
> know how to install your own packages. Eventually when you have been
> using R for a while you will have all sorts of useful functions and
> one day you will think 'What I need is to make a package lucytools to
> keep them together and documented.' This is easier than it looks, I
> wish I had done it much earlier (except obviously I did not call mine
> lucytools).

I just want to add here that you don't need Rtools to install simple 
packages.  If they contain only R code, you can install them without the 
tools.  This advice would likely apply to lucytools (or maybe michaeltools).

My advice doesn't apply to PReMiuM, because that package has C++ code in 
it.  For compiling C, C++, or Fortran code, you need the Rtools, so 
following something like Brian Ripley's or Bill Dunlap's advice is needed.

Duncan Murdoch

>> Thank you also Bill Dunlap for your detailed instructions on how to
>> use RTools, I am sure this will come in handy for me
>> in the future too, and I appreciate you taking the time to write it
>> all out for me.
>> Regards,
>> Lucy Leigh
>>>>> Prof Brian Ripley <ripley at stats.ox.ac.uk> 12/07/2013 6:03 am >>>
>> On 11/07/2013 20:21, David Winsemius wrote:
>>> On Jul 11, 2013, at 1:37 AM, Rolf Turner wrote:
>>>> On 11/07/13 17:57, Prof Brian Ripley wrote:
>>>>> On 11/07/2013 01:22, David Winsemius wrote:
>>>>>> On Jul 10, 2013, at 4:37 PM, Lucy Leigh wrote:
>>>>>>> Hi,
>>>>>>> I have had a look at the manual but it makes no sense to me. I have
>>>>>>> downloaded RTools, and the InnoSetup,
>>>>>>> but I don't understand how to use these to install my package? Am I
>>>>>>> meant to be writing commands
>>>>>>> in R itself, or in these other things I've downloaded?
>>>>>> Since you are clearly out of your league with respect to
>> compiling from source, now is the time to ask (again), why are you
>> not installing the binary package?
>>>>>> At the R command line just type:
>>>>>> install.packages("PReMiuM")  # should default to
>> type="win.binary" and use a CRAN mirror
>>>>> Or as she seems to want to use a later version than on CRAN, to
>> ask again why she does not use winbuilder.
>>>> But the winbuilder web page explicitly says:
>>>>> Please do not upload packages of other maintainers
>> (particularly not without changing the Maintainer field to your own
>> e-mail address, if you have permissions to do that), because the
>> maintainer indicated in the maintainer field of the DESCRIPTION
>> file get response from us. Please do not upload BioConductor
>> packages or CRAN packages. Both BioConductor and CRAN do have build
>> systems. If BioConductor or CRAN packages are not available for
>> Windows, there is a certainly a reason and also this service won't
>> be able to build the package properly.
>>>> So it would appear that it is *not* advisable for Ms. Leigh to
>> use winbuilder.
>>> I read that paragraph as saying that if a specific package , i.e.
>> a particular numbered version, is already on CRAN or the BioC
>> server, then do not duplicate effort. Also do not submit if a CRAN
>> compilation resulted in an error. Especially in light of Prof.
>> Ripley's (reiterated) advice, I did not read it as saying that a
>> pre-release update to an existing package should not be submitted
>> (after suitable alterations of the DESCRIPTION file.)
>> You read correctly, and indeed this is what the R manuals and the rw-FAQ
>> say.  This is all assuming that Lucy does have Silvia's (the author of
>> PReMiuM) permission, but given that she has a unreleased version, that
>> seems eminently reasonable.
>> I am CCing Uwe Ligges (the provider of the winbuilder service) in case
>> he wants to expand the statement.  Although winbuilder does not have a
>> large capacity, new hardware was acquired fairly recently and it is not
>> as hard-pressed as it was.
>> In the particular case of PReMiuM compiling is tricky (it needs the
>> right version of Boost).  And because we've been here before, I know
>> that winbuilder has a suitable version of Boost.
>> An alternative would be for Silvia to submit that version to winbuilder
>> and send the link to Lucy to download.
>> --
>> Brian D. Ripley,                  ripley at stats.ox.ac.uk
>> Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
>> University of Oxford,             Tel:  +44 1865 272861 (self)
>> 1 South Parks Road,                     +44 1865 272866 (PA)
>> Oxford OX1 3TG, UK                Fax:  +44 1865 272595
> Michael Dewey
> info at aghmed.fsnet.co.uk
> http://www.aghmed.fsnet.co.uk/home.html
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

More information about the R-help mailing list