SWIG R with c++11
Locked
 
|
Hi,
is it possible to build the SWIG R package with -std=c++11 ? As far as I can see during the execution of R CMD check R one step * checking whether package ‘QuantLib’ can be installed ... ERROR fails because some hard coded (?) configuration (not related to the one produced with the usual ./configure ... call) is used here, in the log R.Rcheck/00install.out I get make[3]: Entering directory `/home/peter/quantlibpc/QuantLib-SWIG/R/src' g++ -I/usr/share/R/include -DNDEBUG `quantlib-config --cflags` -fpic -O3 -pipe -g -c QuantLib.cpp -o QuantLib.o which fails because of some c++11 elements in my client quantlib code. Thanks a lot Peter ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
|
|
Hi Peter, On 26 October 2013 at 21:34, Peter Caspers wrote: | is it possible to build the SWIG R package with -std=c++11 ? As far | as I can see during the execution of R CMD check R one step | | * checking whether package ‘QuantLib’ can be installed ... ERROR | | fails because some hard coded (?) configuration (not related to the one | produced with the usual ./configure ... call) is used here, in the log | R.Rcheck/00install.out I get | | make[3]: Entering directory `/home/peter/quantlibpc/QuantLib-SWIG/R/src' | g++ -I/usr/share/R/include -DNDEBUG `quantlib-config --cflags` -fpic -O3 -pipe -g -c QuantLib.cpp -o QuantLib.o | | which fails because of some c++11 elements in my client quantlib code. Interesting question. The whole 'R and C++11' nexus is a little toxic (as the CRAN gatekeepers exists on the older standard; some of us are working behind the scenes to change that). By and large, using -std=c++11 should not bite. We use it for a things with Rcpp, but because we cannot (yet!!) upload to CRAN with it, there isn't as much testing for C++11. R itself is in C and does not care. Boost may care (and Boost 2.0 will be guaranteed to work this way). No idea about Swig. I haven't built the QL-Swig bindings in a while. Now, your post doesn't actually show the error. What happens when you simply set appropriate CXXFLAGS as in CXXFLAGS= -g -O3 -Wall -pipe -Wno-unused -pedantic -std=c++11 (I usually do that in ~/.R/Makevars; you can also set it per package in src/Makevars). Can you try and report back? Dirk -- Dirk Eddelbuettel | [hidden email] | http://dirk.eddelbuettel.com ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
|
|
Hi Dirk,
thank you. With CXX=clang++ CXXFLAGS=-g -O3 -Wall -pipe -Wno-unused -pedantic -std=c++11 in ~/R/Makevars make runs smooth on my code. I get 12 warnings of type QuantLib.cpp:552:7: warning: 'register' storage class specifier is deprecated [-Wdeprecated-register] which does not look critical, and 3 warnings * checking DESCRIPTION meta-information ... WARNING Non-standard license specification: QuantLib License Standardizable: FALSE * checking for missing documentation entries ... WARNING Undocumented code objects: ‘ARSCurrency’ ‘ATSCurrency’ ‘AUDCurrency’ ‘AUDLibor’ ‘AUDLibor__SWIG_0’ ‘AUDLibor__SWIG_1’ ‘Actual360’ ‘Actual365Fixed’ ‘Actual365NoLeap’ ‘ActualActual’ ‘ActualActual__SWIG_0’ ... * checking PDF version of manual ... WARNING LaTeX errors when creating PDF version. This typically indicates Rd problems. again not dangerous, are they. Also then running R CMD INSTALL . goes without problems. After that I have a QuantLib.so which I can load from R dyn.load('/usr/local/lib/R/site-library/QuantLib/libs/QuantLib.so') and after source('~/quantlibpc/QuantLib-SWIG/R/R/QuantLib.R') I can succesfully run a few basic tests with QuantLib objects. However there does not seem to be the compiled wrapper QuantLib.RData generated. Was it renamed maybe or do I have to generate it by myself ? Thanks a lot Peter Dirk Eddelbuettel <[hidden email]> writes: > Hi Peter, > > On 26 October 2013 at 21:34, Peter Caspers wrote: > | is it possible to build the SWIG R package with -std=c++11 ? As far > | as I can see during the execution of R CMD check R one step > | > | * checking whether package ‘QuantLib’ can be installed ... ERROR > | > | fails because some hard coded (?) configuration (not related to the one > | produced with the usual ./configure ... call) is used here, in the log > | R.Rcheck/00install.out I get > | > | make[3]: Entering directory `/home/peter/quantlibpc/QuantLib-SWIG/R/src' > | g++ -I/usr/share/R/include -DNDEBUG `quantlib-config --cflags` -fpic -O3 -pipe -g -c QuantLib.cpp -o QuantLib.o > | > | which fails because of some c++11 elements in my client quantlib code. > > Interesting question. The whole 'R and C++11' nexus is a little toxic (as the > CRAN gatekeepers exists on the older standard; some of us are working behind > the scenes to change that). By and large, using -std=c++11 should not bite. > We use it for a things with Rcpp, but because we cannot (yet!!) upload to > CRAN with it, there isn't as much testing for C++11. > > R itself is in C and does not care. Boost may care (and Boost 2.0 will be > guaranteed to work this way). No idea about Swig. I haven't built the QL-Swig > bindings in a while. > > Now, your post doesn't actually show the error. What happens when you simply > set appropriate CXXFLAGS as in > > CXXFLAGS= -g -O3 -Wall -pipe -Wno-unused -pedantic -std=c++11 > > (I usually do that in ~/.R/Makevars; you can also set it per package in > src/Makevars). > > Can you try and report back? > > Dirk ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
|
|
Hi Peter, On 27 October 2013 at 10:45, Peter Caspers wrote: | thank you. With | | CXX=clang++ | CXXFLAGS=-g -O3 -Wall -pipe -Wno-unused -pedantic -std=c++11 | | in ~/R/Makevars make runs smooth on my code. I get 12 warnings of type | | QuantLib.cpp:552:7: warning: 'register' storage class specifier is deprecated [-Wdeprecated-register] | | which does not look critical, and 3 warnings | | * checking DESCRIPTION meta-information ... WARNING | Non-standard license specification: | QuantLib License | Standardizable: FALSE Normal. R stocks a set of licenses (on Debian/Ubuntu in /usr/share/R/share/licenses/) but does not know our one. Which does not matter as we'd never ever push this as a package to CRAN (as we'd need R help page entry for each callable function etc pp) | * checking for missing documentation entries ... WARNING | Undocumented code objects: | ‘ARSCurrency’ ‘ATSCurrency’ ‘AUDCurrency’ ‘AUDLibor’ | ‘AUDLibor__SWIG_0’ ‘AUDLibor__SWIG_1’ ‘Actual360’ ‘Actual365Fixed’ | ‘Actual365NoLeap’ ‘ActualActual’ ‘ActualActual__SWIG_0’ | ... Yep. As expected. | * checking PDF version of manual ... WARNING | LaTeX errors when creating PDF version. | This typically indicates Rd problems. | | again not dangerous, are they. | | Also then running R CMD INSTALL . goes without problems. After that I | have a QuantLib.so which I can load from R | | dyn.load('/usr/local/lib/R/site-library/QuantLib/libs/QuantLib.so') | | and after | | source('~/quantlibpc/QuantLib-SWIG/R/R/QuantLib.R') Nooooooooooooo!!! That is wrong. The changes I made a few months ago allow you to a) create what looks like an R package to R b) install it like an R package c) use it like an R package. In short: R -e 'library(QuantLib); callWhatEverYouWantHere()' and of course also in a longer interactive sesssion. Here QuantLib is the package we got out of QL's SWIG work. Doing it this is d) portable as R knows whether your library ends in .so, .dll or .dynlib e) well understood and a standard in the eosystem you entered: R f) hence _much_ easier on users g) also deals with all the metadata (eg the RData) h) makes R aware of the package too (eg library() sees it) I apologize for having taken years to make the somewhat small changes to all the excellent work Joe had to make it build under SWIG. Once you have a package, it all works much, much better. | I can succesfully run a few basic tests with QuantLib objects. | | However there does not seem to be the compiled wrapper QuantLib.RData | generated. Was it renamed maybe or do I have to generate it by myself ? You do that too as part of the package build and load. Automagically. Please lets keep this thread going. Between you, Klaus, Joe, myself, we shoukld get some mojo back into QuantLib and R. (And yes, I still plan to renovate RQuantLib. Help welcome. Repo on R-Forge, would be happy to move to GH if folks think it helped.) Dirk -- Dirk Eddelbuettel | [hidden email] | http://dirk.eddelbuettel.com ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
|
|
Hi Dirk,
> | > | dyn.load('/usr/local/lib/R/site-library/QuantLib/libs/QuantLib.so') > | > | and after > | > | source('~/quantlibpc/QuantLib-SWIG/R/R/QuantLib.R') > > Nooooooooooooo!!! That is wrong. > don't panic, my fault. I tend to read through readmes too fast, then often missing the lines on the top ... yes, your new way of doing it works very well. > Please lets keep this thread going. Between you, Klaus, Joe, myself, we > shoukld get some mojo back into QuantLib and R. (And yes, I still plan to > renovate RQuantLib. Help welcome. Repo on R-Forge, would be happy to move to > GH if folks think it helped.) sure. Maybe GH would support collaboration better. Don't know R-Forge though. Thanks for you help Peter ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
«
Return to quantlib-dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |