在Docker容器中绘图时Jupyter R崩溃

Question

I am trying to set up Jupyter using Docker. I would like to provide multiple languages kernels (python, scala, R etc.) Now it seems work for others but R. Everytime I execute a 'plot' function, the browser will tell me "The kernel appears to have died. It will restart automatically"

And the error message in log is:

 ***caught segfault ***
 address 0x35da0, cause 'memory not mapped'
 Traceback:
 1: png(tf, width, height, "in", pointsize, bg, res, antialias = antialias)
 2: dev.cb(tf)
 3: repr_recordedplot_generic(obj, ".png", TRUE, function(tf) png(tf, width, height, "in", pointsize, bg, res, antialias = antialias))
 4: repr_png.recordedplot(plotobj)
 5: mime2repr[mime]
 6: doTryCatch(return(expr), name, parentenv, handler)
 7: tryCatchOne(expr, names, parentenv, handlers[[1L]])
 8: tryCatchList(expr, classes, parentenv, handlers)
 9: tryCatch({ formats[[mime]] <- mime2repr[mime]}, error = handle_error)
 10: send_plot(last_recorded_plot)
 11: executor$execute(msg)
 12: handle_shell()
 13: kernel$run()
 14: IRkernel::main()
 aborting ... 
 [I 2016-06-08 19:41:02.360 XX restarter:103] KernelRestarter: restarting kernel (1/5)
 WARNING:root:kernel XXXXXXXXXXXXXXXXXX restarted

If I execute a function of png() or jpeg() from the browser (Google Chrome) first, then plot() function could be executed successfully. I don't know why this happens, what will be changed after running png() or jpeg() . Has anyone met similar issues in either Jupyter or R?

Answer 1

This ended up avoiding using conda while downloading/installing the R packages. Instead, CRAN is used. It seems something is wrong with the compiled packages in conda.