-
Notifications
You must be signed in to change notification settings - Fork 42
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Compilation failed #4
Comments
Have you received any response on this? I am getting the same error. |
@kkdang I have not received a response. |
@slowkow What compute platform did you try this on? I am on Mac OS 10.9.4. I will try one or more Linux distributions to see whether I get a different result. |
Compiled and loaded successfully on Linux 13.04, specifically the Bioconductor AMI. On OSX, didn't work even after installing PCRE, Swig, and Cmake and building from source. |
Hi everyone, apologies. I am surprised this thread is not here: https://github.com/PMBio/peer, which is the main repository. We have heard this before. It looks like the compile error is cause by the clang compiler that is now shipped by default on OSX. We need to update the eigen library to make this work. I have not gotten round to do this yet. In the inters, just stick to GCC and you will be fine. |
I am on Mac OS 10.10.4 (Yosemite). Although I use GCC to compile
I have failed to compile peer. Find full log attached. Any ideas? Thanks!
|
Hi @slowkow and @meritxellop - Were you able to resolve this issue? I encounter the same error and haven't found a solution yet. |
It's two years from the last comment, but with R 3.4, PEER is no longer available as an R package for Mac. To install from the source package you need to compile using GCC, but NOT the system GCC that comes with MacOS. Instead, install a GCC from either MacPorts or Homebrew and use that. Point the CC and CXX variables in .R/Makevars at the resultant gcc/g++ binaries and it should work. Hopefully, this will help someone struggling with this problem who stumbles on this thread searching for an answer. |
Still isn't working. Using R 3.4.3 and GCC/G++ 7.3.0, both from Homebrew, installation still errors out.
Error line is |
have anyone installed PEER in mac successfully? It seems this is an issue for years and no one cared for it. |
PEER uses an ancient version of the Eigen software package, which the
developers seem unlikely to update. So, you should be able to install it
using a similarly ancient version of GCC, eg gcc-4.9. Install GCC 4.9, or
some similarly old version from macports or homebrew and set the CC and CXX
variables in your .R/Makevars to point to it.
…On Thu, Mar 22, 2018 at 4:53 PM, Veera ***@***.***> wrote:
have anyone installed PEER in mac successfully? It seems this is an issue
for years and no one cared for it.
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#4 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AFl0M40OfnYT8IPKAUJYgVeXF8iCEXReks5thB1egaJpZM4C1I_n>
.
|
Installing
seems to get tantalisingly close to success. The message at the end of a long stream of output is
and the namespace fails to load so functions fail. Any help would be greatly appreciated. The PMBio repo is a fork and has Issues off, otherwise I'd raise this there. |
Has anyone installed PEER successfully on Mac OS > 10.11 by now? I have tried nearly all the methods above but still, have the same compiling problem as the first post. |
Hi, I haven't been able to install PEER successfully, after trying for several months. It seems too outdated to use and the maintainers are not maintaining as updates in R have come out. The dependencies of the package are on older packages which makes it difficult to use. I think I will be trying to use SVA instead for my analysis. |
Yes, that's true. I just went for Linux system and used SVA on my Mac instead. |
I have not tried installing PEER on Mac OS, but I recently had success with installing both the Python and R PEER v1.3 packages within a Docker image using Ubuntu 18.04 as the base image. Despite being a different operating system, I had similar issues to the problems mentioned here. In my case, for the R package, the specific GCC compiler version used during install was important. GCC version 5.5.0 worked for me, but GCC 4.x and GCC >=6 did not. For those interested, see the Dockerfile here. EDIT: Linked the Dockerfile instead of copying the code here. |
I was having the same problem but on my Linux machine. I just followed @bryancquach's Dockerfile instructions (lines 86-103) and made sure to use gcc-5 :
select the option for gcc-5, then make peer! |
I followed @KatrionaGoldmann's advice and switched gcc and g++ to version 5 when building from source. Finally, I can get the Python version PEER installed, but not for the R version or Standalone tool. At least, it's better than nothing. To change the default version of the gcc compilers: For the R version, one can probably try to use conda to install r-peer |
Did the version of Cmake make a difference? I am attempting to run a Dockerfile based on @bryancquach 's Dockerfile but running into the error: CMakeFiles/Makefile2:195: recipe for target 'src/CMakeFiles/peerlib.dir/all' failed Has anyone run into this? |
change gcc version and add the Makevars file works for me to do it. |
@wangxinge I tried your method but it returned "no packages specified" for the last command |
Has anyone encountered this issue while installing peer software? $ make |
May I ask that you used the C Make method to install peer? I tried several methods but none worked /sigh |
yes, my R version is 4.0.3, my python version is 2.7.5, my cmake version is 3.20.4, my SWIG version is 2.0.10.
$ cmake ./.. -Wno-dev |
Workaround for Mac computers with Apple silicon (M1 chip, etc.)This was tested in macOS Sonoma (v14.4) with an M1 chip. Apple clang++ was utilized as the C++ compiler. Install clang if unavailable using the command I skipped building the Python package and focused solely on compiling the R package. Native arm64 R app from https://cran.r-project.org/bin/macosx/ did not work. The PEER source package seems to be incompatible with the arm64 architecture. I used conda and installed an old R release, R-3.4.1. I guess most of the x86_64 R distributions should work. 1. Conda environment
2. Clone the repo and edit the source
I edited one line in the R/CMakeLists.txt. Replace the line 40 in R/CMakeLists.txt (https://github.com/PMBio/peer/blob/master/R/CMakeLists.txt#L40) as below. 3. Build the R packageRun the following cmake command. Update the file path to your own conda envs accordingly.
The installation process was successfully replicated by my colleague. The command |
I tried to install the PEER R package and it failed to compile. The instructions are here.
I'm reporting the issue here because the PMBio/peer fork does not have issues enabled.
Notice:
Full log:
The text was updated successfully, but these errors were encountered: