[Rcpp-devel] Wierd compilation error
Hi all, I was experimenting with templates and Rcpp and trying to compile this code: // [[Rcpp::export]] static void readVector(Vector v){ traits::storage_type::type i = v[0]; Rcout << i << std::endl; } I got this error: scratchpad.cpp: In function 'SEXPREC* sourceCpp_3062_readVector(SEXP)': scratchpad.cpp:337:21: error: variable or field '__result' declared void scratchpad.cpp:338:9: error: '__result' was not declared in this scope make: *** [scratchpad.o] Error 1 Also, the line numbers don't make any sense, because the file scratchpad.cpp is only 250 lines long... Do you know what could be wrong? Thanks in advance! Ale -- Alessandro Mammana, PhD Student Max Planck Institute for Molecular Genetics Ihnestraße 63-73 D-14195 Berlin, Germany ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Wierd compilation error
The attributes parser does not know how to handle the static keyword. Works for me without it. You get the line number of the generated file. Use verbose = TRUE to have a better clue at what is wrong. Romain Le 29 janv. 2014 à 12:24, Alessandro Mammana a écrit : > Hi all, > I was experimenting with templates and Rcpp and trying to compile this code: > > > // [[Rcpp::export]] > static void readVector(Vector v){ >traits::storage_type::type i = v[0]; >Rcout << i << std::endl; > } > > I got this error: > > scratchpad.cpp: In function 'SEXPREC* sourceCpp_3062_readVector(SEXP)': > scratchpad.cpp:337:21: error: variable or field '__result' declared void > scratchpad.cpp:338:9: error: '__result' was not declared in this scope > make: *** [scratchpad.o] Error 1 > > Also, the line numbers don't make any sense, because the file > scratchpad.cpp is only 250 lines long... > > Do you know what could be wrong? > > Thanks in advance! > Ale > > -- > Alessandro Mammana, PhD Student > Max Planck Institute for Molecular Genetics > Ihnestraße 63-73 > D-14195 Berlin, Germany > ___ > Rcpp-devel mailing list > Rcpp-devel@lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp and required steps before next release
On 28 January 2014 at 22:10, Dan Tenenbaum wrote: | Thanks. I was actually wondering something slightly different: for the ones that need modification, will they still work with the current Rcpp after being modified, or are the changes not backwards-compatible? I will investigate and I guess I will find out. Yes of course. That is the point of the patches we prepared for CRAN: allow things to work with current CRAN Rcpp 0.10.6, _and_ allow transition to new (to be released) one 0.11.0. We are not trying to intentionally break one or the other, or force things. But without the suggested changes, Rcpp 0.11.0 will NOT get onto CRAN as it will break a few packages. Hope this clarifies. Dirk -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
[Rcpp-devel] Rcpp, Windows and CMake
Hi all, I'm a relative beginner in c/c++ and I'm trying to use c/c++ libraries in an R package. I need it to work in Windows, though, and I'm struggling to get things working. Environment: R 3,0,2 Rtools 3.0 cygwin Native CMake 2.8 cygwin 2.8 The cygwin CMake is convinced that the gcc installed by Rtools doesn't work, because it's passing cygwin paths in the generated Unix Makefiles. The gcc fails the testCCompiler.c check with an error "gcc.exe: fatal error: no input files". If I try to actually compile the file manually, it works as long as I use either relative or native windows paths. and The native CMake refuses to use the Rtools gcc with MinGW Makefiles I have some questions: 1) What would I have to do to be able to use the cygwin mingw compiler instead of the Rtools version? The documentation[http://cran.r-project.org/doc/manuals/R-admin.html#The-MinGW_002dw64-toolchain] states: Users developing packages with Rcpp need to ensure that they use a version built with exactly the same toolchain as their package: the recommendation is to build Rcpp from its sources yourself. Do I simply need to install Rcpp from source using my target compiler? Will that effect my ability to use the binary package on other systems(with CRAN Rcpp)? 2) The RInside package contains several cmake examples. The windows examples seem to imply cygwin, but I am unable to run them as indicated. I get the same gcc error: "no input files". Thanks for any help you can offer, --Jamie Jamie Olson ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
Hi Jamie, On 29 January 2014 at 10:39, Jamie Olson wrote: | Hi all, | I'm a relative beginner in c/c++ and I'm trying to use c/c++ libraries | in an R package. I need it to work in Windows, though, and I'm | struggling to get things working. | | Environment: | R 3,0,2 | Rtools 3.0 | cygwin ^^ Not a good idea, in general, as "R is not supported on Cygwin". Use Rtools. Use something like MSys which gives you bash shell and terminal on top of MinGW. But do make sure Rtools comes first in your path. | Native CMake 2.8 | cygwin 2.8 | | The cygwin CMake is convinced that the gcc installed by Rtools doesn't | work, because it's passing cygwin paths in the generated Unix | Makefiles. The gcc fails the testCCompiler.c check with an error | "gcc.exe: fatal error: no input files". If I try to actually compile | the file manually, it works as long as I use either relative or native | windows paths. | | and | | The native CMake refuses to use the Rtools gcc with MinGW Makefiles CMake should work, outside Cygwin. Folks like Peter (who may chime in) have used CMake with Rcpp/RInside. | I have some questions: | | 1) What would I have to do to be able to use the cygwin mingw compiler | instead of the Rtools version? Port R to Cygwin. Many have tried. Nobody has really succeeded. In short: Don't. | The documentation[http://cran.r-project.org/doc/manuals/R-admin.html#The-MinGW_002dw64-toolchain] | states: | Users developing packages with Rcpp need to ensure that they use a | version built with exactly the same toolchain as their package: the | recommendation is to build Rcpp from its sources yourself. | | Do I simply need to install Rcpp from source using my target compiler? | Will that effect my ability to use the binary package on other | systems(with CRAN Rcpp)? | | 2) The RInside package contains several cmake examples. The windows | examples seem to imply cygwin, but I am unable to run them as | indicated. I get the same gcc error: "no input files". Yes they do, thanks to Peter. But what makes you think it implies Cygwin? I'd be happy to adjust / clarify documentation as needed if you can give me a pointer. Thanks, Dirk | | Thanks for any help you can offer, | --Jamie | | | Jamie Olson | ___ | Rcpp-devel mailing list | Rcpp-devel@lists.r-forge.r-project.org | https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
Thanks Dirk, I'll take a stab with MSys. Just wondering, though, why then are the Rtools commands based on cygwin instead of MSys? I thought that the RInside examples were to be built using Cygwin because they use the "Unix Makefiles" generator. I hadn't yet seen anything to indicate that was possible on windows without cygwin. >From the CMake >wiki[http://www.cmake.org/Wiki/CMake_Generator_Specific_Information#Makefile_generators]: This generator generates Makefiles for make under Unix and also for cygwin (i.e. with Unix paths) Do "Unix Makefiles" work on MSys, too, or should I use "MinGW Makefiles"? Thanks, --Jamie Jamie Olson On Wed, Jan 29, 2014 at 10:56 AM, Dirk Eddelbuettel wrote: > > Hi Jamie, > > On 29 January 2014 at 10:39, Jamie Olson wrote: > | Hi all, > | I'm a relative beginner in c/c++ and I'm trying to use c/c++ libraries > | in an R package. I need it to work in Windows, though, and I'm > | struggling to get things working. > | > | Environment: > | R 3,0,2 > | Rtools 3.0 > | cygwin > ^^ > > Not a good idea, in general, as "R is not supported on Cygwin". > > Use Rtools. > > Use something like MSys which gives you bash shell and terminal on top of > MinGW. But do make sure Rtools comes first in your path. > > | Native CMake 2.8 > | cygwin 2.8 > | > | The cygwin CMake is convinced that the gcc installed by Rtools doesn't > | work, because it's passing cygwin paths in the generated Unix > | Makefiles. The gcc fails the testCCompiler.c check with an error > | "gcc.exe: fatal error: no input files". If I try to actually compile > | the file manually, it works as long as I use either relative or native > | windows paths. > | > | and > | > | The native CMake refuses to use the Rtools gcc with MinGW Makefiles > > CMake should work, outside Cygwin. Folks like Peter (who may chime in) have > used CMake with Rcpp/RInside. > > | I have some questions: > | > | 1) What would I have to do to be able to use the cygwin mingw compiler > | instead of the Rtools version? > > Port R to Cygwin. Many have tried. Nobody has really succeeded. In short: > Don't. > > | The > documentation[http://cran.r-project.org/doc/manuals/R-admin.html#The-MinGW_002dw64-toolchain] > | states: > | Users developing packages with Rcpp need to ensure that they use a > | version built with exactly the same toolchain as their package: the > | recommendation is to build Rcpp from its sources yourself. > | > | Do I simply need to install Rcpp from source using my target compiler? > | Will that effect my ability to use the binary package on other > | systems(with CRAN Rcpp)? > | > | 2) The RInside package contains several cmake examples. The windows > | examples seem to imply cygwin, but I am unable to run them as > | indicated. I get the same gcc error: "no input files". > > Yes they do, thanks to Peter. But what makes you think it implies Cygwin? > I'd be happy to adjust / clarify documentation as needed if you can give me a > pointer. > > Thanks, Dirk > > | > | Thanks for any help you can offer, > | --Jamie > | > | > | Jamie Olson > | ___ > | Rcpp-devel mailing list > | Rcpp-devel@lists.r-forge.r-project.org > | https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel > > -- > Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
On 29 January 2014 at 11:05, Jamie Olson wrote: | Thanks Dirk, I'll take a stab with MSys. Just wondering, though, why | then are the Rtools commands based on cygwin instead of MSys? | | I thought that the RInside examples were to be built using Cygwin | because they use the "Unix Makefiles" generator. I hadn't yet seen | anything to indicate that was possible on windows without cygwin. | >From the CMake wiki[http://www.cmake.org/Wiki/CMake_Generator_Specific_Information#Makefile_generators]: | | This generator generates Makefiles for make under Unix and also for | cygwin (i.e. with Unix paths) | | | Do "Unix Makefiles" work on MSys, too, or should I use "MinGW Makefiles"? Ok, let me speak to the Makefiles shipped with RInside: there is src/Makefile for Linux and OS X, and there is src/Makefile.win for the other OS. Ditto in each of the six inst/examples/* subdirectories. That is what is shipped and supported -- using make and your g++/clang++/... compiler on Linux or OS X, and using make (built for Windows) and your g++ from Rtools. In short, no 'Unix Makefiles' generator anywhere. Maybe this comes from contributed CMake examples. But it seems that this is between you and CMake, and the CMake examples, why useful to those using RIinside with IDEs deploying CMake, are not the documented use case. Maybe get the basic stuff working first. Say 'make -f Makefile.win' in one of the examples directoies. Take it from there. Hope this helps, Dirk -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
Three more things: 1) Is the WIN.readme.txt in examples/eigen/cmake incorrect, then? 2) The Rtools documentation specifies that the Rtools paths should come BEFORE cygwin paths. Is there a similar rule for MSys? 3) Should I continue to just used the MinGW installed with Rtools or should I/can I use a fresh install of MinGW? Thanks, --Jamie Jamie Olson On Wed, Jan 29, 2014 at 11:19 AM, Dirk Eddelbuettel wrote: > > On 29 January 2014 at 11:05, Jamie Olson wrote: > | Thanks Dirk, I'll take a stab with MSys. Just wondering, though, why > | then are the Rtools commands based on cygwin instead of MSys? > | > | I thought that the RInside examples were to be built using Cygwin > | because they use the "Unix Makefiles" generator. I hadn't yet seen > | anything to indicate that was possible on windows without cygwin. > | >From the CMake > wiki[http://www.cmake.org/Wiki/CMake_Generator_Specific_Information#Makefile_generators]: > | > | This generator generates Makefiles for make under Unix and also for > | cygwin (i.e. with Unix paths) > | > | > | Do "Unix Makefiles" work on MSys, too, or should I use "MinGW Makefiles"? > > Ok, let me speak to the Makefiles shipped with RInside: there is src/Makefile > for Linux and OS X, and there is src/Makefile.win for the other OS. > > Ditto in each of the six inst/examples/* subdirectories. That is what is > shipped and supported -- using make and your g++/clang++/... compiler on > Linux or OS X, and using make (built for Windows) and your g++ from Rtools. > > In short, no 'Unix Makefiles' generator anywhere. Maybe this comes from > contributed CMake examples. But it seems that this is between you and CMake, > and the CMake examples, why useful to those using RIinside with IDEs > deploying CMake, are not the documented use case. > > Maybe get the basic stuff working first. Say 'make -f Makefile.win' in one > of the examples directoies. Take it from there. > > Hope this helps, Dirk > > -- > Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
On 29 January 2014 at 11:42, Jamie Olson wrote: | Three more things: | | 1) Is the WIN.readme.txt in examples/eigen/cmake incorrect, then? What part? Setting the CYGWIN env var is to just shut a noisy warning from Rtools off. It does not suggest you should use Cygwin. The rest seems correct to me. | 2) The Rtools documentation specifies that the Rtools paths should | come BEFORE cygwin paths. Is there a similar rule for MSys? MSys is not needed, it is just for you convenience. But yes, I place its part before Cygwin too. | 3) Should I continue to just used the MinGW installed with Rtools or | should I/can I use a fresh install of MinGW? As it stated repeatedly in several places, use the same compiler that R was built on for Windows to build your add-ons such as Rcpp or RInside code. If you know what you're doing you can swap other compilers. One guys apparently happily plugged in a newer MinGW: http://stackoverflow.com/questions/21355022/how-to-use-rcpp-with-tdm-gcc-4-8-1/ I would not recommend that you start there. Keep it simple. Follow the documentation. Things tend to work if you do. Once you know the ropes, do whatever :) Dirk -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
Thanks! The WIN.readme.txt instructs you to run cmake -G "Unix Makefiles" .. It seems that the only cmake on windows that supports "Unix Makefiles" is the cygwin version, and from what you've said it sounds like that can't be used with the native MinGW required for Rcpp. --Jamie Jamie Olson On Wed, Jan 29, 2014 at 12:08 PM, Dirk Eddelbuettel wrote: > > On 29 January 2014 at 11:42, Jamie Olson wrote: > | Three more things: > | > | 1) Is the WIN.readme.txt in examples/eigen/cmake incorrect, then? > > What part? > > Setting the CYGWIN env var is to just shut a noisy warning from Rtools > off. It does not suggest you should use Cygwin. > > The rest seems correct to me. > > | 2) The Rtools documentation specifies that the Rtools paths should > | come BEFORE cygwin paths. Is there a similar rule for MSys? > > MSys is not needed, it is just for you convenience. But yes, I place its part > before Cygwin too. > > | 3) Should I continue to just used the MinGW installed with Rtools or > | should I/can I use a fresh install of MinGW? > > As it stated repeatedly in several places, use the same compiler that R was > built on for Windows to build your add-ons such as Rcpp or RInside code. > > If you know what you're doing you can swap other compilers. One guys > apparently happily plugged in a newer MinGW: > http://stackoverflow.com/questions/21355022/how-to-use-rcpp-with-tdm-gcc-4-8-1/ > I would not recommend that you start there. > > Keep it simple. Follow the documentation. Things tend to work if you do. > Once you know the ropes, do whatever :) > > Dirk > > -- > Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
On 29 January 2014 at 13:03, Jamie Olson wrote: | Thanks! | | The WIN.readme.txt instructs you to run | | cmake -G "Unix Makefiles" .. | | It seems that the only cmake on windows that supports "Unix Makefiles" | is the cygwin version, and from what you've said it sounds like that | can't be used with the native MinGW required for Rcpp. Oh. Now I see. Well, you may have to talk to Peter who wrote the README. Dirk -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
On Wed, Jan 29, 2014 at 9:39 AM, Jamie Olson wrote: > Hi all, > I'm a relative beginner in c/c++ and I'm trying to use c/c++ libraries > in an R package. I need it to work in Windows, though, and I'm > struggling to get things working. > > --- >8--- > It might help to state what c/c++ libraries you're trying to get up and running. There is some documentation in RTools (IIRC) on setting up an RTOOLS_LOCAL encironment that was pretty straight forward. I had success building the libraries for GMP, MPFR, and such using mingw-64 then placing the resulting files in an RToolsLocal directory. If you go that route you should be able to use cmake to build and then install into your LOCAL path. Or, are you trying to have this cmake be part of a package that you are creating? The native CMake refuses to use the Rtools gcc with MinGW Makefiles > > This doesn't surprise me at all, as the RTools toolchain requires that the environment is setup, which it is when calling _from_ R. You'd have to figure out how to do that manually if you want to use that g++ from else where (say msys). It can be done, but as you mention... --->8--- > > The documentation[ > http://cran.r-project.org/doc/manuals/R-admin.html#The-MinGW_002dw64-toolchain > ] > states: > Users developing packages with Rcpp need to ensure that they use a > version built with exactly the same toolchain as their package: the > recommendation is to build Rcpp from its sources yourself. > > Do I simply need to install Rcpp from source using my target compiler? > Will that effect my ability to use the binary package on other > systems(with CRAN Rcpp)? > > You _can_ install Rcpp from source on windows using a recent msys environment with a recent g++ ( I did last December ). But there be dragons down that road! You have to make sure that you get all of the settings correct and all of that jazz, which is a major pain; and then you are still going to have problems unless you can figure out the proper procedure for getting the unwinding info correct, and the cross boundary communication between libraries correct or you will end up freezing R with every error your code produces. I would've kept with a g++ 4.8.2 or clang 3.4 if it wasn't for the fact that I couldn't get any graceful exception handling. I know I make it sound bad. That is on purpose! You said you are a relative beginner... I'm someone who likes to 'just dive in'; and I can tell you that I dove in right where you are standing and you're gonna your head if you do it that way. --->8--- > > Thanks for any help you can offer, > --Jamie > > > Jamie Olson > ___ > Rcpp-devel mailing list > Rcpp-devel@lists.r-forge.r-project.org > https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel > -- Sincerely, Thell ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
I'm trying to build the Avro-C library[http://avro.apache.org/docs/1.7.5/api/c/index.html]. There is a README for building in windows[http://svn.apache.org/repos/asf/avro/trunk/lang/c/README.maintaining_win32.txt] but it only has instructions for Visual Studio. I've recently been wondering if that was intentional, since now cmake-generated MSYS Makefiles are giving errors. Are you suggesting that I build the library with MinGW(in cygwin/linux?) and then just copy it to link to when building the package with Rcpp? Jamie Olson On Wed, Jan 29, 2014 at 2:07 PM, Thell Fowler wrote: > > > > On Wed, Jan 29, 2014 at 9:39 AM, Jamie Olson > wrote: >> >> Hi all, >> I'm a relative beginner in c/c++ and I'm trying to use c/c++ libraries >> in an R package. I need it to work in Windows, though, and I'm >> struggling to get things working. >> >> --- >8--- > > > It might help to state what c/c++ libraries you're trying to get up and > running. There is some documentation in RTools (IIRC) on setting up an > RTOOLS_LOCAL encironment that was pretty straight forward. I had success > building the libraries for GMP, MPFR, and such using mingw-64 then placing > the resulting files in an RToolsLocal directory. If you go that route you > should be able to use cmake to build and then install into your LOCAL path. > > Or, are you trying to have this cmake be part of a package that you are > creating? > > >> The native CMake refuses to use the Rtools gcc with MinGW Makefiles >> > > This doesn't surprise me at all, as the RTools toolchain requires that the > environment is setup, which it is when calling _from_ R. You'd have to > figure out how to do that manually if you want to use that g++ from else > where (say msys). It can be done, but as you mention... > >> --->8--- >> >> >> The >> documentation[http://cran.r-project.org/doc/manuals/R-admin.html#The-MinGW_002dw64-toolchain] >> states: >> Users developing packages with Rcpp need to ensure that they use a >> version built with exactly the same toolchain as their package: the >> recommendation is to build Rcpp from its sources yourself. >> >> Do I simply need to install Rcpp from source using my target compiler? >> Will that effect my ability to use the binary package on other >> systems(with CRAN Rcpp)? >> > > You _can_ install Rcpp from source on windows using a recent msys > environment with a recent g++ ( I did last December ). But there be dragons > down that road! You have to make sure that you get all of the settings > correct and all of that jazz, which is a major pain; and then you are still > going to have problems unless you can figure out the proper procedure for > getting the unwinding info correct, and the cross boundary communication > between libraries correct or you will end up freezing R with every error > your code produces. I would've kept with a g++ 4.8.2 or clang 3.4 if it > wasn't for the fact that I couldn't get any graceful exception handling. > > I know I make it sound bad. That is on purpose! You said you are a > relative beginner... I'm someone who likes to 'just dive in'; and I can > tell you that I dove in right where you are standing and you're gonna your > head if you do it that way. > > >> --->8--- >> >> >> Thanks for any help you can offer, >> --Jamie >> >> >> Jamie Olson >> >> ___ >> Rcpp-devel mailing list >> Rcpp-devel@lists.r-forge.r-project.org >> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel > > > > > -- > Sincerely, > Thell ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
On 29 January 2014 at 15:07, Jamie Olson wrote: | I'm trying to build the Avro-C | library[http://avro.apache.org/docs/1.7.5/api/c/index.html]. There is | a README for building in | windows[http://svn.apache.org/repos/asf/avro/trunk/lang/c/README.maintaining_win32.txt] | but it only has instructions for Visual Studio. I've recently been | wondering if that was intentional, since now cmake-generated MSYS | Makefiles are giving errors. | | Are you suggesting that I build the library with MinGW(in | cygwin/linux?) and then just copy it to link to when building the | package with Rcpp? In a nutshell, yes. If you want to / need to package Avro as a library it may be trickier. At a first pass try to build a static library, then link your Rcpp package to that. Dirk -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp and required steps before next release
On 29 January 2014 at 12:06, Dan Tenenbaum wrote: | Yes, thanks. I took a look at https://github.com/kevinushey/RcppBiocChecks and it looks like Bioconductor packages are in good shape. Thanks Kevin for checking this, and Steffen for fixing mzR and Rdisop. | | That document does refer to several packages that are not Bioconductor packages: | | CDM | sirt | TAM | geiger Note the date. Kevin's page is a little dusty (~ 5 days). I am trying to keep https://github.com/RcppCore/rcpp-logs/blob/master/status/status-20140129.txt more current (which includes changing the url which may be bad style -- but I am keeping the url of all other files in that repo). CDM, sirt, and TAM are indeed on the list of the now 12 packages still needing a patch. The other one, geiger, morphed to works as-is with some other work. And Kevin also confirmed to me that the four CRAN packages I didn't touch (as they depended on BioC packages) also test fine -- so we are now in the clear for all of BioC. I still need to reflect that update from Kevin. But we're getting there -- and it is really excellent how folks are helping. Four patched packages are already on CRAN, plus three which got there late last week, plus the handful I was involved with, ... | Anyway, I don't think I need to take further action on this, but if anyone needs me to do or test anything, just let me know. Noted, thanks. Dirk -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
1) When doing that, there won't be any problem using a different version of MinGW? 2) I'm guessing I'd want to do this in linux, to avoid including the cygwin dlls? --Jamie Jamie Olson On Wed, Jan 29, 2014 at 3:13 PM, Dirk Eddelbuettel wrote: > > On 29 January 2014 at 15:07, Jamie Olson wrote: > | I'm trying to build the Avro-C > | library[http://avro.apache.org/docs/1.7.5/api/c/index.html]. There is > | a README for building in > | > windows[http://svn.apache.org/repos/asf/avro/trunk/lang/c/README.maintaining_win32.txt] > | but it only has instructions for Visual Studio. I've recently been > | wondering if that was intentional, since now cmake-generated MSYS > | Makefiles are giving errors. > | > | Are you suggesting that I build the library with MinGW(in > | cygwin/linux?) and then just copy it to link to when building the > | package with Rcpp? > > In a nutshell, yes. > > If you want to / need to package Avro as a library it may be trickier. At a > first pass try to build a static library, then link your Rcpp package to that. > > Dirk > > -- > Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
On 29 January 2014 at 15:22, Jamie Olson wrote: | 1) When doing that, there won't be any problem using a different | version of MinGW? I thought we had been over this before: yes, mixing is a problem, so no, do not do this. Build R and Rcpp and your preferred libraries with the same version. I even built Qt and what not that way years ago. | 2) I'm guessing I'd want to do this in linux, to avoid including the | cygwin dlls? I am not following. Were you not constrained by being on Windows? Dirk -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
Sorry, I think I just got confused. I meant cross-compiling from linux with mingw since cygwin's cmake is giving me trouble, but it's probably more effort than it's worth to get the same compiler. Thanks for all your help! --Jamie Jamie Olson On Wed, Jan 29, 2014 at 3:44 PM, Dirk Eddelbuettel wrote: > > On 29 January 2014 at 15:22, Jamie Olson wrote: > | 1) When doing that, there won't be any problem using a different > | version of MinGW? > > I thought we had been over this before: yes, mixing is a problem, so no, do > not do this. Build R and Rcpp and your preferred libraries with the same > version. > > I even built Qt and what not that way years ago. > > | 2) I'm guessing I'd want to do this in linux, to avoid including the > | cygwin dlls? > > I am not following. Were you not constrained by being on Windows? > > Dirk > > -- > Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
Re: [Rcpp-devel] Rcpp, Windows and CMake
On 29 January 2014 at 16:18, Jamie Olson wrote: | Sorry, I think I just got confused. I meant cross-compiling from | linux with mingw since cygwin's cmake is giving me trouble, but it's | probably more effort than it's worth to get the same compiler. Yes. Cross-compiling can work, but some of the version of gcc are more finicky than other. Eg Jeroen used the different mingw cross-compiler version on Ubuntu to build a Windows library of Protocol Buffers which he then made available to CRAN / win-builder. So now we have Windows builds of RProtoBuf too. But he had to try three different ones. As I recall the 4.7.* variant worked, the others lead to errors once used on Windows. But you'd have to ask him. We are a little off-topic now as far as 'how do I use Rcpp' goes. | Thanks for all your help! My pleasure. Dirk -- Dirk Eddelbuettel | e...@debian.org | http://dirk.eddelbuettel.com ___ Rcpp-devel mailing list Rcpp-devel@lists.r-forge.r-project.org https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel