本文章已经设置了最低额度的付费阅读，如果您觉得文章对您有用，且手头宽裕，欢迎请作者喝杯热茶。本文章付费部分内容并不影响您对文章的理解，只是作者对付费阅读的一次尝试和态度，感谢支持。

背景

R 语言中的 venneuler 是一个用于计算并显示韦恩图和欧拉图的包，这个包在 CRAN 上的最后一个版本是 1.1.0，发布于 2011-08-10，它是一个基于 rJava 依赖的 R 包。

  1. vd <- venneuler(c(A=0.3, B=0.3, C=1.1, "A&B"=0.1, "A&C"=0.2, "B&C"=0.1 ,"A&B&C"=0.1))
  2. plot(vd)
  4. # same as c(A=1, `A&B&C`=1, C=1)
  5. m <- data.frame(elements=c("1","2","2","2","3"), sets=c("A","A","B","C","C"))
  6. v <- venneuler(m)
  7. plot(v)
  9. m <- as.matrix(data.frame(A=c(1.5, 0.2, 0.4, 0, 0),
  10.                           B=c(0 , 0.2, 0 , 1, 0),
  11.                           C=c(0 , 0 , 0.3, 0, 1)))
  12. # without weights
  13. v <- venneuler(m > 0)
  14. plot(v)
  16. # with weights
  17. v <- venneuler(m)
  18. plot(v)

在 venneuler 包，作者引入了一个的用于描述集合交集的向量，这一点后来也被 UpSetR 所借鉴，即 UpSetR 中的 fromExpression 。 UpSetR 接受三种类型的数据输入：

  • 表格数据，即 R 语言里面的数据框。行表示元素，列表示数据集分配和额外信息。
  • 元素名的集合( fromList )。
  • venneuler 包引入的用于描述集合交集的向量 ( fromExpression）。

早在18年6月的时候，我公众号上写过一篇关于 UpSetR 的学习笔记《UpSetR：多数据集绘图可视化处理利器》，提到过一下 venneuler ，当时也没怎么留意，直到前不久有个读者在使用 UpSetR 的时候给我了一个截图。

image.png

  • 其实，这个截图中的 fromExpression 拼写错了，如果 UpSetR 安装好，又把 fromExpression 拼对，就可以解决导入集合交集向量数据的问题。
  • rJava 、 venneuler 包的安装并没有想象中的那么好安装，尤其是 venneuler 。

问题

使用 conda install r-venneuler 安装完 venneuler 包后， library(venneuler) 加载时出现动态库异常。

  1. > library(venneuler)
  2. Loading required package: rJava
  3. Error occurred during initialization of VM
  4. Unable to load native library: /usr/local/software/miniconda3/libjava.so: cannot open shared object file: No such file or directory
  1. $ R CMD javareconf
  2. Java interpreter : /usr/local/software/miniconda3/bin/java
  3. Java version     : 1.8.0_192
  4. Java home path   : /usr/local/software/miniconda3/jre
  5. Java compiler    : /usr/local/software/miniconda3/bin/javac
  6. Java headers gen.: /usr/local/software/miniconda3/bin/javah
  7. Java archive tool: /usr/local/software/miniconda3/bin/jar
  9. trying to compile and link a JNI program
  10. detected JNI cpp flags    : -I$(JAVA_HOME)/../include -I$(JAVA_HOME)/../include/linux
  11. detected JNI linker flags : -L$(JAVA_HOME)/lib/amd64/server -ljvm
  12. x86_64-conda_cos6-linux-gnu-cc -I"/usr/local/software/miniconda3/lib/R/include" -DNDEBUG -I/usr/local/software/miniconda3/jre/../include -I/usr/local/software/miniconda3/jre/../include/linux  -DNDEBUG -D_FORTIFY_SOURCE=2 -O2 -isystem /usr/local/software/miniconda3/include -I/usr/local/software/miniconda3/include -Wl,-rpath-link,/usr/local/software/miniconda3/lib  -fpic  -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -ffunction-sections -pipe -isystem /usr/local/software/miniconda3/include -fdebug-prefix-map=/home/conda/feedstock_root/build_artifacts/r-base_1576190804673/work=/usr/local/src/conda/r-base-3.6.2 -fdebug-prefix-map=/usr/local/software/miniconda3=/usr/local/src/conda-prefix  -c conftest.c -o conftest.o
  13. x86_64-conda_cos6-linux-gnu-cc -shared -L/usr/local/software/miniconda3/lib/R/lib -Wl,-O2 -Wl,--sort-common -Wl,--as-needed -Wl,-z,relro -Wl,-z,now -Wl,--disable-new-dtags -Wl,--gc-sections -Wl,-rpath,/usr/local/software/miniconda3/lib -Wl,-rpath-link,/usr/local/software/miniconda3/lib -L/usr/local/software/miniconda3/lib -Wl,-rpath-link,/usr/local/software/miniconda3/lib -o conftest.so conftest.o -L/usr/local/software/miniconda3/jre/lib/amd64/server -ljvm -L/usr/local/software/miniconda3/lib/R/lib -lR
  16. JAVA_HOME        : /usr/local/software/miniconda3/jre
  17. Java library path: $(JAVA_HOME)/lib/amd64/server
  18. JNI cpp flags    : -I$(JAVA_HOME)/../include -I$(JAVA_HOME)/../include/linux
  19. JNI linker flags : -L$(JAVA_HOME)/lib/amd64/server -ljvm
  20. Updating Java configuration in /usr/local/software/miniconda3/lib/R
  21. Done.
  23. $ cd /usr/local/software/miniconda3
  24. $ find ./ -name libjava.so
  25. ./pkgs/openjdk-8.0.192-h516909a_1004/jre/lib/amd64/libjava.so
  26. ./jre/lib/amd64/libjava.s

即使使用 LD_LIBRARY_PATH 添加完目标动态库后，依然无法避免 “java/lang/NoClassDefFoundError: java/lang/Object“ 的异常。

  1. $ export LD_LIBRARY_PATH=/usr/local/software/miniconda3/jre/lib/amd64:$LD_LIBRARY_PATH
  2. $ R
  4. R version 3.6.2 (2019-12-12) -- "Dark and Stormy Night"
  5. Copyright (C) 2019 The R Foundation for Statistical Computing
  6. Platform: x86_64-conda_cos6-linux-gnu (64-bit)
  8. R is free software and comes with ABSOLUTELY NO WARRANTY.
  9. You are welcome to redistribute it under certain conditions.
  10. Type 'license()' or 'licence()' for distribution details.
  12.   Natural language support but running in an English locale
  14. R is a collaborative project with many contributors.
  15. Type 'contributors()' for more information and
  16. 'citation()' on how to cite R or R packages in publications.
  18. Type 'demo()' for some demos, 'help()' for on-line help, or
  19. 'help.start()' for an HTML browser interface to help.
  20. Type 'q()' to quit R.
  22. > library(venneuler)
  23. Loading required package: rJava
  24. Error occurred during initialization of VM
  25. java/lang/NoClassDefFoundError: java/lang/Object

使用 conda create -n r-3.6.2 r-base=3.6.2 创建的新环境中发现， conda install -r r-venneuler 会同时执行 openjdk 、 r-rjava 、 r-venneuler 三个包的安装，并解决相关的依赖，但即使安装过程没有任何问题，在使用 library(venneuler) 后上面提到的问题依然会出现。

conda 环境中 openjdk 、 r-rjava 、 r-venneuler 都可以安装成功，但就是用不了，这是我当前遇到的问题，暂时没找到解决方法。

解决

抛弃了 conda 后，用最原始的方法，终于可以解决这个问题，下面是记录。

1. 安装二进制的 java

在 http://www.oracle.com/technetwork/java/javase/downloads/java-archive-downloads-javase7-521261.html 下载 jdk-7u80-linux-x64.tar.gz，解压缩。

  1. $ tar zvxf jdk-7u80-linux-x64.tar.gz
  2. $ mv jdk1.7.0_80 /usr/local/software

2. Java 环境配置

把下面的内容添加到 ~/.bashrc 最后， source ~/.bashrc 执行环境变量更新。

  1. export PATH=/usr/local/software/jdk1.7.0_80/bin:$PATH
  2. export JAVA_HOME=/usr/local/software/jdk1.7.0_80
  3. export JRE_HOME=/usr/local/software/jdk1.7.0_80/jre
  4. export LD_LIBRARY_PATH=/usr/local/software/jdk1.7.0_80/lib/amd64:/usr/local/software/jdk1.7.0_80/jre/lib/amd64:$LD_LIBRARY_PATH

3. 更新 R 的 java 配置

如果没有更新 R 语言的 java 支持配置，直接执行 isntall.packages(rJava) 可能会引发以下问题：

  1. checking whether JNI programs run... configure: error: Unable to run a simple JNI program. 
  2. Make sure you have configured R with Java support (see R documentation) and check config.log for failure reason.

所以，请使用下面的命令更新 R 语言的 java 支持配置：

  1. $ /usr/local/software/miniconda3/bin/R CMD javareconf
  2. Java interpreter : /usr/local/software/jdk1.7.0_80/jre/bin/java
  3. Java version     : 1.7.0_80
  4. Java home path   : /usr/local/software/jdk1.7.0_80
  5. Java compiler    : /usr/local/software/jdk1.7.0_80/bin/javac
  6. Java headers gen.: /usr/local/software/jdk1.7.0_80/bin/javah
  7. Java archive tool: /usr/local/software/jdk1.7.0_80/bin/jar
  9. trying to compile and link a JNI program
  10. detected JNI cpp flags    : -I$(JAVA_HOME)/include -I$(JAVA_HOME)/include/linux
  11. detected JNI linker flags : -L$(JAVA_HOME)/jre/lib/amd64/server -ljvm
  12. x86_64-conda_cos6-linux-gnu-cc -I"/usr/local/software/miniconda3/lib/R/include" -DNDEBUG -I/usr/local/software/jdk1.7.0_80/include -I/usr/local/software/jdk1.7.0_80/include/linux  -DNDEBUG -D_FORTIFY_SOURCE=2 -O2 -isystem /usr/local/software/miniconda3/include -I/usr/local/software/miniconda3/include -Wl,-rpath-link,/usr/local/software/miniconda3/lib  -fpic  -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -ffunction-sections -pipe -isystem /usr/local/software/miniconda3/include -fdebug-prefix-map=/home/conda/feedstock_root/build_artifacts/r-base_1576190804673/work=/usr/local/src/conda/r-base-3.6.2 -fdebug-prefix-map=/usr/local/software/miniconda3=/usr/local/src/conda-prefix  -c conftest.c -o conftest.o
  13. x86_64-conda_cos6-linux-gnu-cc -shared -L/usr/local/software/miniconda3/lib/R/lib -Wl,-O2 -Wl,--sort-common -Wl,--as-needed -Wl,-z,relro -Wl,-z,now -Wl,--disable-new-dtags -Wl,--gc-sections -Wl,-rpath,/usr/local/software/miniconda3/lib -Wl,-rpath-link,/usr/local/software/miniconda3/lib -L/usr/local/software/miniconda3/lib -Wl,-rpath-link,/usr/local/software/miniconda3/lib -o conftest.so conftest.o -L/usr/local/software/jdk1.7.0_80/jre/lib/amd64/server -ljvm -L/usr/local/software/miniconda3/lib/R/lib -lR
  16. JAVA_HOME        : /usr/local/software/jdk1.7.0_80
  17. Java library path: $(JAVA_HOME)/jre/lib/amd64/server
  18. JNI cpp flags    : -I$(JAVA_HOME)/include -I$(JAVA_HOME)/include/linux
  19. JNI linker flags : -L$(JAVA_HOME)/jre/lib/amd64/server -ljvm
  20. Updating Java configuration in /usr/local/software/miniconda3/lib/R
  21. Done.

4. 安装 rJava 和 venneuler

上面的步骤完成后，在 R 中使用 install.packages("rJava") 和 install.packages("venneuler") ，发现一切正常。

后话