Hi Dirk, Thank you very much, your answers are always helpful and point me to the right direction. And I also hope the C++11 proposal ( https://stat.ethz.ch/pipermail/r-devel/2013-October/067677.html) will be approved soon. :-)
Simon On 11/03/2013 09:54 PM, Dirk Eddelbuettel wrote: > On 3 November 2013 at 11:35, Simon wrote: > | Hi, > | > | Recently, I made an R package that used the C++ library Boost.Thread > (http://www.boost.org/doc/libs/1_54_0/doc/html/thread.html) for > multithreading. Previously, I have posted a question at stackoverflow > (http://stackoverflow.com/questions/19651954/is-it-possible-to-build-an-r-package-which-use-rcpp-and-boost-thread-on-http), > and I also asked at rcpp-devel > (http://lists.r-forge.r-project.org/pipermail/rcpp-devel/2013-November/006777.html). > Dirk Eddelbuettel suggested me ask here. After some tries on win-builder > and some research on CRAN, there are still some doubts: > > Thanks for posting here, and for including code. I tried to answer both your > questions before. You are now providing more and more context which helps. > > | 1) For my package to be accepted by CRAN, I am not so sure if it is > acceptable to let users to install Boost library themselves? If not, are > there any solutions? I found there are some third party external software or > libraries used by other R packages, such as OpenMP and gtk, some of which > have corresponding R packages, but there are little discuss about Boost. > | > | (I now know there are some other ways to host my package, but I think CRAN > is more straight forward, though it may be stricter.) > > This question is really several questions: > > 1a) If I write a package using library xyz, do users need xyz to compile my > package? > > Yes of course. > > 1b) If I write a package using library xyz, do _windows_ users need xyz to > _use_ my package? > > No as CRAN provides Windows binary packages. > > 1c) What about the other OSs? > > It depends. If users need to compile first, then yes. But Boost is very > standard on OS X and Linux. > > So yes, it is quite acceptable to depend on Boost. > > > | 2) As I said in my question at stackoverflow > (http://stackoverflow.com/questions/19651954/is-it-possible-to-build-an-r-package-which-use-rcpp-and-boost-thread-on-http), > I have tried MAKE variable BOOSTLIB on win-builder > (http://win-builder.r-project.org/), It is OK for header-only Boost > libraries. I am not sure whether it is OK for separately-compiled libraries > such as Boost.Thread, after many fails. Is it possible? Are there any > special variables or path for Boost.Thread? May I suppose that it is > acceptable for 1) if there are, or at least for header-only Boost libraies? > > I do not know if BOOSTLIB on win-builder and CRAN also points to a library > you can use. And I cannot think of a package using it. Maybe Uwe Ligges will > be able to help you here. > > | (There are an R package called BH provides header-only Boost libraries, but > BOOSTLIB make me imagine something else.) > > BOOSTLIB predates the creation of our BH package by several years. > > Dirk > > [[alternative HTML version deleted]] ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
