irkernel / irkernel Goto Github PK

…or deprecate and hide it from the docs.

R has a notion of the “default graphics device”, i.e. getOption('device'), which is either a string containing a function name or a function.

we could leverage this in two ways:

1. use it with custom options:

   dev <- getOption('device')
   if (!is.function(dev)) {
       dev <- get(dev)
   }
   do.call(dev, as.list(options('jupyter.plot')))

2. require the user to override the defaults by setting it to a function:

   #our initialization
   options(device = function(file.name, width = 7, height = 7,...) {
       png(file.name, width = width, height = height, ...)
   })
   
   …
   
   dev <- getOption('device')
   if (!is.function(dev)) {
       dev <- get(dev)
   }
   dev()

the only problem: we want to allow different mime types (PNG, JPG, SVG) for plots: if it’s a function, we can’t derive the type from the option.

The following code runs ok in the command line but not in the notebook:

library(caret)
data(iris)
model <- train(Species~., data=iris, method="nb")

On the notebook it fails with:

Error in frameTypes(env): not a proper evaluation environment

After loading the Bioconductor org.Sc.sgd.db package (and its dependencies)

library(org.Sc.sgd.db)
as.list(org.Sc.sgdGENENAME)

works in a standard console but fails in IPython notebook with the error

Error in as.list.default(org.Sc.sgdGENENAME): no method for coercing this S4 class to a vector

In the standard console showMethods("as.list") returns

Function: as.list (package base)
x="AgiAnnDbMap"
x="AnnDbBimap"
  (inherited from: x="Bimap")
x="ANY"
x="Bimap"
x="CompressedAtomicList"
x="CompressedNormalIRangesList"
x="FlatBimap"
x="GoAnnDbBimap"
x="GOTermsAnnDbBimap"
x="Hits"
x="IpiAnnDbMap"
x="List"
x="Rle"
x="SimpleList"
x="Vector"

while in IPython it returns

Function: as.list (package base)
x="AgiAnnDbMap"
x="ANY"
x="Bimap"
x="CompressedAtomicList"
x="CompressedNormalIRangesList"
x="FlatBimap"
x="GoAnnDbBimap"
x="GOTermsAnnDbBimap"
x="Hits"
x="IpiAnnDbMap"
x="List"
x="Rle"
x="SimpleList"
x="Vector"

So it seems that for some reason the as.list method for the AnnDbBimap doesn't get inherited from Bimap in IPython.

IRkernel::installspec()
Traceback (most recent call last):
File "C:\Anaconda\Scripts\ipython-script.py", line 5, in
sys.exit(start_ipython())
File "C:\Anaconda\lib\site-packages\IPython__init__.py", line 120, in start_ipython
return launch_new_instance(argv=argv, _kwargs)
File "C:\Anaconda\lib\site-packages\IPython\config\application.py", line 574, in launch_instance
app.start()
File "C:\Anaconda\lib\site-packages\IPython\terminal\ipapp.py", line 367, in start
return self.subapp.start()
File "C:\Anaconda\lib\site-packages\IPython\kernel\kernelspecapp.py", line 142, in start
return self.subapp.start()
File "C:\Anaconda\lib\site-packages\IPython\kernel\kernelspecapp.py", line 83, in start
replace=self.replace,
File "C:\Anaconda\lib\site-packages\IPython\kernel\kernelspec.py", line 195, in install_kernel_spec
shutil.copytree(source_dir, destination)
File "C:\Anaconda\lib\shutil.py", line 171, in copytree
names = os.listdir(src)
WindowsError: [Error 3] The system cannot find the path specified: u'C:/Program_.'

If you suspect this is an IPython bug, please report it at:
https://github.com/ipython/ipython/issues
or send an email to the mailing list at [email protected]

You can print a more detailed traceback right now with "%tb", or use "%debug"
to interactively debug it.

Extra-detailed tracebacks for bug-reporting purposes can be enabled via:
c.Application.verbose_crash=True

Warning message:
running command 'ipython kernelspec install --replace --name ir --user C:/Program Files/R/R-3.1.3/library/IRkernel/kernelspec' had status 1

The following code (regression with instrumental variable)

library("lfe")
N <- 1e6
df <- data.frame(
  v1 =  sample(5, N, TRUE),                          # int in range [1,5]
  v2 =  sample(1e6, N, TRUE),                        # int in range [1,1e6]
  v3 =  sample(round(runif(100, max = 100), 4), N, TRUE) # numeric e.g. 23.5749
)
felm(v1 ~ 1 | 0 | (v2 ~ v3) , df)

outputs the error

Error in eval(expr, envir, enclos): object 'v2(fit)' not found

Would you know what is happening?

I'll send some of this in as a pull request.

Hi I've installed IRkernel using bits and pieces I've found here and there, and I seem to have hit a snag. Before I describe the problem in detail, a little background. I run windows, which I know makes this harder. In particular, I built rzmq for windows using this guide:

http://blogs.msdn.com/b/gpalem/archive/2013/02/06/building-zeromq-for-r-rzmq-with-visual-studio-express-full.aspx

and got around another problem using this tip:

http://ihrke.github.io/jupyter.html

So far, things seem to work fine except for one issue: opening an iPython notebook and choosing an 'R' kernal, I run library("vars"), load some data (my_data), and then I run test <- ur.df(my_data, type = "trend", lags = 2). When I run test, I get, as expected

############################################################### 
# Augmented Dickey-Fuller Test Unit Root / Cointegration Test # 
############################################################### 

The value of the test statistic is: -7.9808 

However, when I run summary(test)

I get

Length  Class   Mode 
     1  ur.df     S4 

I should be getting a far more detailed result regarding the Dickey-Fuller unit root test. Digging a little further, I tried

class(test)

and got

[1] "ur.df"
attr(,"package")
[1] "urca"

and class(summary(test)) gives me

[1] "summaryDefault" "table"  

(I believe it should be in the class sumurca according to the documentation for the packages vars and urca)

Also, showMethods("summary") gives

Function: summary (package base)
object="ANY"
object="ca.jo"
object="ca.po"
object="cajo.test"
object="lm"
    (inherited from: object="ANY")
object="ur.df"
object="ur.ers"
object="ur.kpss"
object="ur.pp"
object="ur.sp"
object="ur.za"

When I try any of this in the command line R (or Rgui), it works as expected. Any idea what's going on? As you may have noticed, I'm no expert on any of this, but I've done a fair bit of googling, and I'm stumped. I also have no idea if this is unique to one package.

My setup is windows 7 (64-bit), R 3.1.2, iPython 3. I'm sure there is some other info you need to make a sensible recommendation and, given some guidance, I'm happy to provided it. Running sessionInfo(), I get

R version 3.1.2 (2014-10-31)
Platform: x86_64-w64-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=English_United States.1252 
[2] LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] vars_1.5-2        lmtest_0.9-33     urca_1.2-8        strucchange_1.5-0
[5] sandwich_2.3-2    zoo_1.7-11        MASS_7.3-39      

loaded via a namespace (and not attached):
 [1] base64_1.1      digest_0.6.8    evaluate_0.5.5  grid_3.1.2     
 [5] IRdisplay_0.1   IRkernel_0.2    jsonlite_0.9.14 lattice_0.20-30
 [9] rzmq_0.7.7      stringr_0.6.2   uuid_0.1-1     

Thank you for any thoughts.

Cheers,

DS

I would like to switch from R -e to Rscript to run a bunch of code before calling IRkernel::main. Could you explain where the value of '{connection_file}' gets set when initiating ipyr? Is this inline template code that gets replaced by ipython? Or does IRkernel::main literally need to be called with this string?

The kernel dies as it's started repeatedly with an error like this, before eventually succeeding:

Error in rawToChar(receive.socket(socket, unserialize = FALSE)) : 
  embedded nul in string: '\0ȎpO'
Calls: <Anonymous> ... <Anonymous> -> handle_shell -> receive.multipart -> rawToChar
Execution halted

The string it displays is different each time - presumably it succeeds when the data doesn't have a null byte in.

Using install.packages results in an error from rjson

Error in toJSON(msg$content): unable to convert R type 3 to JSON

Note that the jsonlite package is an alternative to rjson which should be a bit more well behaved. Although I have no idea what R type 3 is.

When running:

> library(devtools)
> install_github("takluyver/IRdisplay")
Downloading github repo takluyver/IRdisplay@master
Error in system(full, intern = quiet, ignore.stderr = quiet, ...) : 
  error in running command

The current implementation buffers printed output in a textConnection() until the user code has finished executing, then sends a single stream message with all the text. So you see no output until the cell finishes.

If we can implement an object suitable to pass to sink, output can be displayed as it is printed.

2014-07-05 10:34:21.973 [NotebookApp] Kernel started: 4293024b-0a7a-4f74-91c7-fa
69678481da

R version 3.1.0 (2014-04-10) -- "Spring Dance"
Copyright (C) 2014 The R Foundation for Statistical Computing
Platform: x86_64-w64-mingw32/x64 (64-bit)

Loading 'meta' package (version 3.6-0).
> IRkernel::main('C:\Users\outwen\.ipython\profile_default\security\kernel-42930
24b-0a7a-4f74-91c7-fa69678481da.json')
: "'C:\U"'\U'hex

2014-07-05 10:34:24.971 [NotebookApp] KernelRestarter: restarting kernel (1/5)
WARNING:root:kernel 4293024b-0a7a-4f74-91c7-fa69678481da restarted

R version 3.1.0 (2014-04-10) -- "Spring Dance"
Copyright (C) 2014 The R Foundation for Statistical Computing
Platform: x86_64-w64-mingw32/x64 (64-bit)

Loading 'meta' package (version 3.6-0).
> IRkernel::main('C:\Users\outwen\.ipython\profile_default\security\kernel-42930
24b-0a7a-4f74-91c7-fa69678481da.json')
: "'C:\U"'\U'hex

2014-07-05 10:34:27.987 [NotebookApp] KernelRestarter: restarting kernel (2/5)
WARNING:root:kernel 4293024b-0a7a-4f74-91c7-fa69678481da restarted

R version 3.1.0 (2014-04-10) -- "Spring Dance"
Copyright (C) 2014 The R Foundation for Statistical Computing
Platform: x86_64-w64-mingw32/x64 (64-bit)


Loading 'meta' package (version 3.6-0).
> IRkernel::main('C:\Users\outwen\.ipython\profile_default\security\kernel-42930
24b-0a7a-4f74-91c7-fa69678481da.json')
: "'C:\U"'\U'hex

2014-07-05 10:34:31.230 [NotebookApp] KernelRestarter: restarting kernel (3/5)
WARNING:root:kernel 4293024b-0a7a-4f74-91c7-fa69678481da restarted

R version 3.1.0 (2014-04-10) -- "Spring Dance"
Copyright (C) 2014 The R Foundation for Statistical Computing
Platform: x86_64-w64-mingw32/x64 (64-bit)

Loading 'meta' package (version 3.6-0).
> IRkernel::main('C:\Users\outwen\.ipython\profile_default\security\kernel-42930
24b-0a7a-4f74-91c7-fa69678481da.json')
: "'C:\U"'\U'hex

2014-07-05 10:34:34.438 [NotebookApp] KernelRestarter: restarting kernel (4/5)
WARNING:root:kernel 4293024b-0a7a-4f74-91c7-fa69678481da restarted


> IRkernel::main('C:\Users\outwen\.ipython\profile_default\security\kernel-42930
24b-0a7a-4f74-91c7-fa69678481da.json')
: "'C:\U"'\U'hex

2014-07-05 10:34:44.163 [NotebookApp] WARNING | KernelRestarter: restart failed
2014-07-05 10:34:44.164 [NotebookApp] WARNING | Kernel 4293024b-0a7a-4f74-91c7-f
a69678481da died, removing from map.
ERROR:root:kernel 4293024b-0a7a-4f74-91c7-fa69678481da restarted failed!

Can IRkernel work in Windows? I just get this information when running it.
Thanks.

it doesn’t save everything. the docs say it’s the same as

save(list = ls(all.names = TRUE), file = ".RData", envir = .GlobalEnv)

something here goes wrong. is it .GlobalEnv or ls()?

Right now there are two related repos, IR_kernel and IRdisplay, that are named with a slightly different convention. I know I am being completely OCD but it would make more sense to have them both with the underscore or without. Awesome work though, I am going to have a student start to use this ASAP.

evaluate creates a new graphics device for each line, and this seems to create Rplots{n}.pdf files in the CWD.

@takluyver @karthik I know this is still work in progress, but I couldn't resist trying it out. I followed the instructions in your README, but was greeted by the message that the kernel was restarting

Dead kernel
The kernel has died, and the automatic restart has failed. It is possible the kernel cannot be restarted. If you are not able to restart the kernel, you will still be able to save the notebook, but running code will no longer work until the notebook is reopened.

Digging further, I see the following message on my console

Error: could not find function "poll.socket"
Execution halted
2014-03-05 09:43:27.892 [NotebookApp] WARNING | KernelRestarter: restart failed

I am using the version of the rzmq package on CRAN.

Any thoughts/pointers on how I might debug this.

install.package('base64enc') is necessary to make the installation instructions work

Plots can be displayed inline as PNG or SVG, using the pyout or display_data messages.

The implementation of IPython's rmagic is a useful reference - it sets up a temporary directory for figures, and then displays any figures created after the R code has run:

https://bitbucket.org/lgautier/rpy2/src/343a654fc54bb4965e92130bdd7768d799d945b9/rpy/ipython/rmagic.py?at=version_2.4.x#cl-669

currently ?something outputs the raw help ANSI like _�D_�e_�s_�c_�r_�i_�p_�t_�i_�o_�n: (underscore backspace letter)

the Rd help should be converted using Rd2HTML, not Rd2TXT

f <- function(){message("ok")}
f()

returns

ok # in R
simpleMessage in message("ok"): ok # in IPython

It can be cumbersome when the code inside message is long. For instance,

left_join(DT,DT1)

returns

Joining by: c("id", "v1") # R
simpleMessage in message("Joining by: ", capture.output(dput(by))): Joining by: c("id", "v1") #Ipython

Is there a simple way to simplify the Ipython output? Sorry - I have no idea how to change that (I could not even find the word "simple" in the repository).

Kernel gets stuck when using help:

?glm
help(glm)

From a discussion with @ramnathv.

Note that being in IRkernel does not necessarily imply that the code is running in the notebook - there are multiple frontends speaking the same protocol, and they may even be connected to the kernel at the same time. But we can reliably indicate that we're in the kernel.

Hi,

Looking more closely the error messages when installing on windows, I notice this one

> install_github('armstrtw/rzmq', pull=8, ref=NULL)
Erreur : Please install jsonlite package

it seems to suggest an added step in installation

install_github("jsonlite")
(or ?)
install_github("jeroenooms/jsonlite")

(apparently, it's IRKernel, that wants it )

importFrom(jsonlite,fromJSON)

so ...

install_github("jeroenooms/jsonlite")

.... and crash

`      -----------------------------------
* installing *source* package 'jsonlite' ...
Warning: running command 'sh ./configure.win' had status 127
ERROR: configuration failed for package 'jsonlite'
* removing 'C:/Users/famille/AppData/Local/Temp/Rtmp0kndNo/Rinstd2c505d1de/jsonl
ite'
      -----------------------------------
ERROR: package installation failed
Erreur : Command failed (1)

Just in case, I posted the question on jsonlite project
jeroen/jsonlite#28

For cells with "cell_type": "code", you would expect "language": "R" but it the .ipynb file says "language": "python".

@karthik and @sckott, both using Macs, both had an issue where code appeared to run, but no output was visible. However, @ramnathv, also on a Mac, doesn't see it, and it works for a couple of people on Linux. Similar to an earlier, fixed issue (see discussion on #2).

No idea what could be causing this at the moment, but please comment with any ideas, debugging, or best of all, solutions.

Is there a way to create a profile or something else so that one can start the native R kernel using something like

$ ipython notebook --profile=R

When I type
> help()
in IPython 2.2.0, the help documentation pops up from the lower part of the screen.
See the screenshot

but there is no pop-up help message for IPython 3.0.0 dev.
I checked the terminal output and the help message also does not show up there.

There is similar buggy behavior for

contributors()

It would be nice to prepare a package for rzmq and irkernel, so that windows user can install them without setting up a mingw environment. Or upload them to cran :-)

I did get it to install the 64 bit version:

• install.packages('RCurl') and library(devtools)
• compile zeromg with a mingw by following this howto: http://zeromq.org/docs:windows-installations#toc2. I changed Makefile to remove perf and test from SUBDIRS to make compiling faster. This builds a 64bit libzmq.dll, but no 32 bit... I couldn't get it to compile using the rtools (setting the path to c:/rtools/bin and using C:\data\downloads\zeromq-4.0.5\zeromq-4.0.5>sh configure --prefix=C:/rtools/zeromq build_alias=x86-mingw32 "CFLAGS=-m32" "CXXFLAGS=-m32" "LDFLAGS=-m32") and also not the 32bit version with the above mingw (passing in "CFLAGS=-m32" "CXXFLAGS=-m32" "LDFLAGS=-m32" to configure...)
• clone armstrtw/rzmq and git submodule init/update to get inst\cppzmq\zmq.hpp. Also copied zmq.h from the compiled zeromq folder into that dir, because I didn't know how to find the usr/local/include folder in rtools...
• copied libzmq.dll together with libgcc_s_sjlj-1.dll and libstdc++-6.dll into a lot of subfolders of C:\Rtools\gcc-4.6.3 (*\bin, *\lib; not sure which was the correct one in the end) until the following worked:
• in rzmq folder: C:/PROGRA~1/R/R-31~1.2/bin/x64/R --vanilla CMD INSTALL %CD% --library="C:/portabel/Rlibs" --no-multiarch
• install_github("takluyver/IRdisplay")
• clone takluyver/IRkernel and in irkernel again: C:/PROGRA~1/R/R-31~1.2/bin/x64/R --vanilla CMD INSTALL %CD% --library="C:/portabel/Rlibs" --no-multiarch
• run C:/PROGRA~1/R/R-31~1.2/bin/x64/R and then IRkernel::installspec()
• running a v3 ipython notebook (from git yesterday) with a r kernel didn't work at first, as the R started by ipython notebook was 32bit and zmq wasn't installed in that version (see above...), so modified the .ipython\kernels\ir\kernel.json to read {"argv": ["R","--arch", "x64", "-e","IRkernel::main()","--args","{connection_file}"], (-> added "--arch", "x64",). I couldn't find a way to set this the default R, but that's probably just again missing google foo on my side :-( Next time I will simple install only 64bit R and not both.... [EDIT: it's probably enough to setx R_ARCH x64]

And yay: there was an R kernel and I can run R code in the notebook :-)

The above was brought to you by 6 hours poking around and a lot of trial and error in the middle of the night, numerous google searches and now not enough sleep but neverless a happy R kernel user :-) Thanks a lot for your work!

Is there a way to produce more output from kernel crashes. Executing code, and then Jupyter says the kernel has crashed with no other output. If we could understand some troubleshooting steps we could report back to the project on the issue.

Thanks!

Are you planning uploading this to CRAN?

If not, can you at least tag a release? I'm trying to build a conda package from this, and I'd like to have a stable release point other than just "master" to build off of.

Syntax highlighting would be a nice enhancement.

By the way, why do "$" signs get red blocks around (in the current version)?

I'd like to use the function [ provided by data.table. However, I obtain

library(data.table)
DT <- data.table(id=1:100)
DT[1:10]
#> Error in `[.data.frame`(x, i): undefined columns selected

Maybe this is related to this SO question

I tried to test the new autocompletion feature, but I wasn't able to obtain good results.

I (re)installed according to the README (maybe here is when I broke my installation):

# Need RCurl for install_github
install.packages('RCurl')
library(devtools)
install_github('armstrtw/rzmq')
install_github("takluyver/IRdisplay")
install_github("takluyver/IRkernel")

# Only if you have IPython 3 or above installed:
IRkernel::installspec()

And after opening a notebook, I got this:

And some errors in the console:

 [I 18:43:34.578 NotebookApp] Adapting to protocol v4.0 for kernel 2da601ae-8408-4be6-b6e3-b61743ed876c
 [1] "Got unhandled msg_type:" "object_info_request"  

If a type ta or data. and hit Tab, then I get good results.

By the way, are we going to get tooltips as part of the implementation? Because right now Shift-Tab has no effect.

Hi,

It's really nice that IRkernel lets you display graphs inline. But is there a way to change the display width and height?

I'm running the NMF package in R and with parallelization and I found that It broken in the IRkernel with ta invalid value of 'seed' [in call to 'nextRNGStream'] error. I tested this in R and in the IPython Notebook with %%R magics and it works. I'm not sure but can this be a ZMQ/rzmq/pyzmq issue?

So, in R, if you define an object of data.table, subsetting is different from data.frame, so for example:

library(data.table)
DT <- data.table(x=rnorm(9), y=repc(("a", "b", "c"), each=3), z=rnorm(9))
DT[, c(2,3)] should return
2 3
but DT[,list(x,y)] can be used to access 2nd & 3rd columns. However, in ipython-Rkernel, DT behaves like data.frame rather than a data.table. Therefore, DT[, list(x,y)] gives an error

So, normally in R, when you use plot() function, a new figure is constructed, and following lines() or points() or legend() functions will add something to that figure! However, in RKernel for ipython it is different as each new function will continue to a new figure. An example is provided here:

http://nbviewer.ipython.org/github/mirjalil/DataScience/blob/master/R.Programs/RKernel_plot_bug.ipynb

Is there a good way to run single lines without writing them in a cell? Existing guides use %qtconsole which seems specific to python

I'm not sure what the root of the problem is but here is the minimal example I have:

for (i in 1:5) {
    cat('=')
    Sys.sleep(.2)
}

the above waits 1 second and then prints 5 equals signs. This is apposed to the following code:

cat('=')
Sys.sleep(.2)
cat('=')
Sys.sleep(.2)
cat('=')
Sys.sleep(.2)
cat('=')
Sys.sleep(.2)
cat('=')
Sys.sleep(.2)

that prints an equal sign every .2 seconds which is what I would expect from both examples. I played around with julia and python and they do what I expect in for loops.

FWIW I am running the ipython from master and R is version 3.1.2.

Handle a complete_request to make tab completion work.

I remember you had pushed a fix for this, but for some reason I am still getting stray Rplots.pdf files in my working directory. Any thoughts?

I updated some packages on my ubuntu machine which previously was working, and now it broke. When trying to plot, the notebook displays

Error in base_display("image/png", b64data, metadata): unused argument (metadata)

Looking at IRdisplay it seems that the third argument of base_display was deprecated. But where is base_display called from?

Hi all,

when I try to create an S4 class in the notebook, e.g.,

setClass("circle", representation(r="numeric"))

it gives me the following error:

Error in assign(mname, def, where): cannot add bindings to a locked environment

This problem is easily fixed if one specifies the environment in the setClass function:

setClass("circle", representation(r="numeric"), where=.GlobalEnv)

But my question now is: why is the environment in IPython notebook not .GlobalEnv? Why is it locked?

I have a second (perhaps related?) problem after loading the Bioconductor org.Sc.sgd.db package (and its dependencies)

library(org.Sc.sgd.db)
as.list(org.Sc.sgdGENENAME)

works in a standard console but fails in IPython notebook with the error

Error in as.list.default(org.Sc.sgdGENENAME): no method for coercing this S4 class to a vector

I don't know what's going on since showMerhods("as.list") shows that there is a method for the S4 class AnnDbBimap.

Any help is appreciated.

System: OS X Yosemite, R: @3.1.2_0+accelerate+cairo+gfortran48+recommended+x11

> install_github('armstrtw/rzmq')
Downloading github repo armstrtw/rzmq@master
Installing rzmq
'/opt/local/Library/Frameworks/R.framework/Resources/bin/R' --vanilla CMD  \
  INSTALL  \
  '/private/var/folders/rv/wstf88ns4g16sqgg62lsz_pw0000gn/T/Rtmp5NeM5l/devtools182a058697683/armstrtw-rzmq-320891d'  \
  --library='/Users/jiaweihe/Library/R/3.1/library' --install-tests 
 installing *source* package ‘rzmq’ ...
** libs
/usr/bin/clang++ -std=c++11 -I/opt/local/Library/Frameworks/R.framework/Resources/include -DNDEBUG -I../inst/cppzmq -I/opt/local/include    -fPIC  -pipe -Os -arch x86_64 -stdlib=libc++ -c interface.cpp -o interface.o
interface.cpp:24:10: **fatal error: 'zmq.hpp' file not found**
#include <zmq.hpp>
         ^
1 error generated.
make: *** [interface.o] Error 1
ERROR: compilation failed for package ‘rzmq’
 removing ‘/Users/jiaweihe/Library/R/3.1/library/rzmq’

Error: Command failed (1)

like shown here, ipython/jupyter has display hooks that can be registered with display formatters (e.g. png, html)

it would be nice if IRDisplay and IRKernel would work together so that

1. we define a S3 or S4 generic (e.g. jupyterdisplay.default)
2. IRDisplay defines it for some classes (e.g. jupyterdisplay.matrix)
3. users can override it

Am curious if there is at any point any way/plan to convert R scripts to iPython Notebooks with the R kernel. Similar to how one can transform a .py file into a .ipynb file like so:

import IPython.nbformat.current as nbf
nb = nbf.read(open('test.py', 'r'), 'py')
nbf.write(nb, open('test.ipynb', 'w'), 'ipynb')

Is this farfetched? Seems like it would be a great way to increase adoption.

Hi,

I am on MacOs 10.9, I'm not able to install "rzmq",
I got this error message

* installing to library ‘/Library/Frameworks/R.framework/Versions/3.1/Resources/library’
* installing *source* package ‘rzmq’ ...
** package ‘rzmq’ successfully unpacked and MD5 sums checked
** libs
clang++ -I/Library/Frameworks/R.framework/Resources/include -DNDEBUG  -I/usr/local/include -I/usr/local/include/freetype2 -I/opt/X11/include    -fPIC  -Wall -mtune=core2 -g -O2  -c interface.cpp -o interface.o
interface.cpp:22:10: fatal error: 'zmq.hpp' file not found
#include <zmq.hpp>
         ^
1 error generated.
make: *** [interface.o] Error 1
ERROR: compilation failed for package ‘rzmq’
* removing ‘/Library/Frameworks/R.framework/Versions/3.1/Resources/library/rzmq’

thanks

The command substitute does not behave in the same way in a normal R shell and in a Jupyter notebook.

For example:

x <- 5
substitute(y + x)

returns y + x in an R shell, but y + 5 in an R notebook.

This is probably due to the fact that the environment in an R notebook is not the R_GlobalEnv.

This could cause some issues when NSE is involved.

It seems as if STDOUT and STDERR are piped to the server console, see e.g. the R startup message, or output from install.packages. These streams often contain important information for the user. Is there some way to display these in the client?