Re: [R-SIG-Mac] Multiple R versions: set Current via symbolic link: permissions?
Corrected with the '-m32' flag for each of the compilers. On 1/26/12 5:22 PM, Peter Carr wrote: I get compiler errors: ld: duplicate symbol _Rf_mkChar in CommandLineArgs.o and CConverters.o However, I did not explicitly use 32-bit. Is this a compiler flag? I will look into it. On 1/26/12 5:15 PM, Simon Urbanek wrote: 2) specifically to R 2.5, which I need, I get compiler errors. > You'll have to share much more detail. Note that such an ancient version of R may not be buildable on your system without SDKs for older OS X versions (and note that recent Xcode tools default to 64-bit which was not supported back then so you may need to use 32-bit). ___ R-SIG-Mac mailing list R-SIG-Mac@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-mac
Re: [R-SIG-Mac] Multiple R versions: set Current via symbolic link: permissions?
I get compiler errors: ld: duplicate symbol _Rf_mkChar in CommandLineArgs.o and CConverters.o However, I did not explicitly use 32-bit. Is this a compiler flag? I will look into it. On 1/26/12 5:15 PM, Simon Urbanek wrote: 2) specifically to R 2.5, which I need, I get compiler errors. > You'll have to share much more detail. Note that such an ancient version of R may not be buildable on your system without SDKs for older OS X versions (and note that recent Xcode tools default to 64-bit which was not supported back then so you may need to use 32-bit). ___ R-SIG-Mac mailing list R-SIG-Mac@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-mac
Re: [R-SIG-Mac] Multiple R versions: set Current via symbolic link: permissions?
On Jan 26, 2012, at 4:44 PM, Peter Carr wrote: > That thread does help and it provides enough information to solve most of my > problem. However, it is not clear to me how to change the R_HOME directory so > that it is not in the /Library/Frameworks/R.framework directory. This is > important to me because I am providing a tool which will automatically > install R and I don't want my tool to replace an existing version of R. > > I tried building from source, and immediately ran into two problems: >1) I don't know enough to know how to configure my build so that it > doesn't use the Frameworks location You don't need to use the framework version of R if you don't need the GUI. If your copy will be "private" you can use --disable-R-framework and set rhome at install time to wherever you want (or alternatively you can set the path of the framework if you need it). Note that you'll have to re-compile all packages as well. >2) specifically to R 2.5, which I need, I get compiler errors. > You'll have to share much more detail. Note that such an ancient version of R may not be buildable on your system without SDKs for older OS X versions (and note that recent Xcode tools default to 64-bit which was not supported back then so you may need to use 32-bit). > I'd like to simply take apart the framework build I get when install from the > R-2.5.1.dmg. It is possible, albeit convoluted. The easy part of to modify R_HOME - you simply do that in the R script. The harder part is that all binaries have absolute paths to libraries inside the framework. You will have to go through every single binary (package, module, executable) and change all paths to your new location using install_name_tool. To be honest, given how ancient your R is I wouldn't even bother and just keep it in its 2.5 folder since chances that it will clash on user's machine are close to zero. But if you choose to tackle this, you can take inspiration from this script (it uses @rpath which is a bit more flexible but requires more support from binaries): https://svn.r-project.org/R-dev-web/trunk/QA/Simon/R-build/rpath.sh An alternative fix is to use DYLD_LIBRARY_PATH to prepend your new location of libraries. It may or may not work, depending on what you use from R. It is a bit fragile but is fairly easy to do so you can test whether it works for your particular setup of using R. > However, I'd also like this to work, even when I move this to a different > location on my file system. I appreciate some guidance on this last step. > That is actually close to impossible unless you are ok with giving up package installation. The problem is that a) binary package have full paths encoded and b) if you compile a package it will get full path and they need a path to compile. However, if you are shipping a "fixed" R that will not be modified by adding packages, you can use the same idea as above with the special @executable_path prefix. You'll need to perform some gymnastics to infer R_HOME in the R script since it will by dynamic, but it's possible -- the following patch will give you an idea: https://svn.r-project.org/R-dev-web/trunk/QA/Simon/R-build/R-rpath-diff The executive summary is that what you want is possible, but not really easy. Cheers, Simon ___ R-SIG-Mac mailing list R-SIG-Mac@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-mac
Re: [R-SIG-Mac] Multiple R versions: set Current via symbolic link: permissions?
That thread does help and it provides enough information to solve most of my problem. However, it is not clear to me how to change the R_HOME directory so that it is not in the /Library/Frameworks/R.framework directory. This is important to me because I am providing a tool which will automatically install R and I don't want my tool to replace an existing version of R. I tried building from source, and immediately ran into two problems: 1) I don't know enough to know how to configure my build so that it doesn't use the Frameworks location 2) specifically to R 2.5, which I need, I get compiler errors. I'd like to simply take apart the framework build I get when install from the R-2.5.1.dmg. However, I'd also like this to work, even when I move this to a different location on my file system. I appreciate some guidance on this last step. Thanks, Peter ___ R-SIG-Mac mailing list R-SIG-Mac@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-mac
Re: [R-SIG-Mac] Visualizing coefficients
Hi Jarrett, Take a look at the effects package, in particular ?effect, which includes methods for mer and lme objects. (I'm not sure why you'd post this to the r-sig-mac list.) Best, John John Fox Sen. William McMaster Prof. of Social Statistics Department of Sociology McMaster University Hamilton, Ontario, Canada http://socserv.mcmaster.ca/jfox/ On Thu, 26 Jan 2012 10:52:52 -0500 Jarrett Byrnes wrote: > Before I re-invent a well built wheel, has anyone on this list put together a > good set of functions for visualizing net effects and the variation in the > estimates for a fitter lmer or glmer model? I realize that this can be done > to some extent via simulation and then putting the results into coda or > R2WinBUGS, but, has anyone gone about it via a different route? I'm also > curious to track down other resources for visualizing the results of fitted > mer objects. > > -Jarrett > > > > > > Jarrett Byrnes > Postdoctoral Fellow > National Center for Ecological Analysis and Synthesis > 735 State Street, Suite 300 > Santa Barbara, CA 93101 > > http://jarrettbyrnes.info > > > [[alternative HTML version deleted]] > > ___ > R-SIG-Mac mailing list > R-SIG-Mac@r-project.org > https://stat.ethz.ch/mailman/listinfo/r-sig-mac ___ R-SIG-Mac mailing list R-SIG-Mac@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-mac
[R-SIG-Mac] Visualizing coefficients
Before I re-invent a well built wheel, has anyone on this list put together a good set of functions for visualizing net effects and the variation in the estimates for a fitter lmer or glmer model? I realize that this can be done to some extent via simulation and then putting the results into coda or R2WinBUGS, but, has anyone gone about it via a different route? I'm also curious to track down other resources for visualizing the results of fitted mer objects. -Jarrett Jarrett Byrnes Postdoctoral Fellow National Center for Ecological Analysis and Synthesis 735 State Street, Suite 300 Santa Barbara, CA 93101 http://jarrettbyrnes.info [[alternative HTML version deleted]] ___ R-SIG-Mac mailing list R-SIG-Mac@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-mac
