Trouble installing on Ubuntu 12.04

5,939 views
Skip to first unread message

Gene Leynes

unread,
Aug 1, 2012, 2:48:53 PM8/1/12
to cra...@googlegroups.com
Hello, I was hoping to get help with an ubuntu install.

Since the R conference I've been excited to try this, only to be thwarted by my un-cool Windows OS, and then my paltry 32 bit CPUs (which is so old that it can't even run a 64 bit VM).

Then I realized that I could do it in the cloud!  

[Fast forward two weeks and though 9 Ubuntu instance setup attempts] 

Now I have an Ubuntu instance running in the cloud that I set up only to run cranvas and R Studio, but I still get an error during the installation.

Warning in fun(libname, pkgname) :
  Unable to load library icui18n "Cannot load library icui18n: (libicui18n.so.48: cannot open shared object file: No such file or directory)"
Warning in fun(libname, pkgname) : qtbase: cannot connect to X server
ERROR: loading failed
* removing '/home/ubuntu/R/x86_64-pc-linux-gnu-library/2.15/qtbase'

I've followed the directions on the git wiki, so can someone offer advice?  I've tried a few different things without luck.  Googling icui18n didn't help either.

I'm happy to set up a different instance type (e.g. Linux) for testing.  

Thanks!!

Gene

Instance details:
AMI: ubuntu/images/ebs/ubuntu-precise-12.04-amd64-server-20120616 (ami-82fa58eb)

Installation Messages: 


Yihui Xie

unread,
Aug 4, 2012, 5:16:09 PM8/4/12
to cra...@googlegroups.com
I googled for a while, and it seems you need to install libicu48
(although I do not know why), so you probably can try:

sudo apt-get install libicu48

Regards,
Yihui
--
Yihui Xie <xiey...@gmail.com>
Phone: 515-294-2465 Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA

Gene Leynes

unread,
Aug 4, 2012, 5:29:04 PM8/4/12
to cra...@googlegroups.com
Thanks, I'll try that

I wasn't sure where to post the question

By the way, I LOVE using markdown

Sent from my iPhone

Yihui Xie

unread,
Aug 4, 2012, 5:52:36 PM8/4/12
to cra...@googlegroups.com
Either this mailing list or Github is OK. I was just back from JSM,
hence the late reply.

Regards,
Yihui
--
Yihui Xie <xiey...@gmail.com>
Phone: 515-294-2465 Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA


Gene Leynes

unread,
Aug 6, 2012, 3:45:09 PM8/6/12
to cra...@googlegroups.com
I ran this line:
sudo apt-get install libicu48  

but I'm still getting the exact same error.

Could it be a problem with the installation path of my packages?  I didn't realize this at first, but I my R library is not read write enabled.  I install packages to my personal R library.

Yihui Xie

unread,
Aug 6, 2012, 4:14:38 PM8/6/12
to cra...@googlegroups.com
I install R packages to my HOME directory as well. That should not
make any difference. How did you install R? through apt-get install?

Regards,
Yihui
--
Yihui Xie <xiey...@gmail.com>
Phone: 515-294-2465 Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA


Gene Leynes

unread,
Aug 6, 2012, 5:33:01 PM8/6/12
to cra...@googlegroups.com
Yes, I used apt-get install.  I don't know another way in Linux / Ubuntu! 

I think that maybe the problem is that I'm using R in R Studio.  I'm trying again, but this time at the R command line though SSH.  

I'll let you know how it goes (later tonight, it takes a while!)

Yihui Xie

unread,
Aug 6, 2012, 5:34:33 PM8/6/12
to cra...@googlegroups.com
Actually apt-get install is the preferred approach. Yes, please try R
in the terminal directly.

Regards,
Yihui
--
Yihui Xie <xiey...@gmail.com>
Phone: 515-294-2465 Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA


Gene Leynes

unread,
Aug 6, 2012, 6:54:50 PM8/6/12
to cra...@googlegroups.com
I get a different error when I do the install in R in Terminal.  

I think that me and cranvas were not meant to be...  But I think I'll try it in a fresh instance, maybe something got messed up.

Scanning dependencies of target smokeqt
make[3]: Leaving directory `/tmp/Rtmpal8f6z/R.INSTALL1ec649a2716b/qtbase/kdebindings-build'
make[3]: Entering directory `/tmp/Rtmpal8f6z/R.INSTALL1ec649a2716b/qtbase/kdebindings-build'
[ 67%] Building CXX object smoke/qt/CMakeFiles/smokeqt.dir/smokedata.cpp.o
g++: internal compiler error: Killed (program cc1plus)
Please submit a full bug report,
with preprocessed source if appropriate.
See <file:///usr/share/doc/gcc-4.6/README.Bugs> for instructions.
make[3]: *** [smoke/qt/CMakeFiles/smokeqt.dir/smokedata.cpp.o] Error 4
make[3]: Leaving directory `/tmp/Rtmpal8f6z/R.INSTALL1ec649a2716b/qtbase/kdebindings-build'
make[2]: *** [smoke/qt/CMakeFiles/smokeqt.dir/all] Error 2
make[2]: Leaving directory `/tmp/Rtmpal8f6z/R.INSTALL1ec649a2716b/qtbase/kdebindings-build'
make[1]: *** [all] Error 2
make[1]: Leaving directory `/tmp/Rtmpal8f6z/R.INSTALL1ec649a2716b/qtbase/kdebindings-build'
make: *** [all] Error 2
ERROR: compilation failed for package 'qtbase'
* removing '/usr/local/lib/R/site-library/qtbase'
Error: Command failed (1)

Gene Leynes

unread,
Aug 6, 2012, 7:02:01 PM8/6/12
to cra...@googlegroups.com
What is your sessionInfo()  ?

Yihui Xie

unread,
Aug 6, 2012, 10:29:52 PM8/6/12
to cra...@googlegroups.com
It is just a normal R 2.15.1 for 64bit Ubuntu:

> sessionInfo()
R version 2.15.1 (2012-06-22)
Platform: x86_64-pc-linux-gnu (64-bit)

locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=C LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

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


Regards,
Yihui
--
Yihui Xie <xiey...@gmail.com>
Phone: 515-294-2465 Web: http://yihui.name
Department of Statistics, Iowa State University
2215 Snedecor Hall, Ames, IA


On Mon, Aug 6, 2012 at 7:02 PM, Gene Leynes <gle...@gmail.com> wrote:
> What is your sessionInfo() ?

Susan Vander Plas

unread,
Aug 30, 2012, 2:28:01 PM8/30/12
to cra...@googlegroups.com
Try installing qt4-devtools, qt4-qmake, qt4-designer, qt4-doc, libqt4-dev, libqt4-opengl-dev. You may not need them all, but installing them got rid of that error message for me.

Gene Leynes

unread,
Apr 26, 2013, 11:15:36 AM4/26/13
to cra...@googlegroups.com
I was going through my bookmarks, and noticed this bookmark from last year and thought I would update.

I got cranvas "working" this week. I'd like to share a little about what I've learned.  Cranvas and it's dependencies use the computer's hardware in a very direct way, especially video card.  So, that has to be in place and and working in the way it expects for things to work.

Since the conference I've purchased a beastly computer with a huge GPU, and I've been anxious to try out Cranvas.... but it still didn't work out of the box in a virtual machine.  It also didn't work when I installed Ubuntu 12.04... or 12.10, or 13....

After a great deal of searching, it turns out that my computer uses something called Optimus which "intelligently" switches between the power hungry Nvidia GPU and the power efficient Intel GPU.  This causes problems with the drivers on Linux, and the default Noveau drivers don't work with the graphics card.  They don't actually recognize the graphics card at all, but it does actually use the less powerful Intel GPU (I think) and doesn't touch the Nvidia card.  Incidentally, the 13.04 does seem to know that it's using the Intel GPU, but it has other problems.

Updating the Nvidia card's drivers using just about any method you can find on the Nvidia website will cause big problems.  Basically the desktop environment gets hosed in one way or another, and you can only do command line stuff.  Sometimes I found ways to repair this, but reinstalling the OS was easier.

For most applications the answer turns out to be something called Bumblebee, which is developed by these really awesome people in the UK.  http://bumblebee-project.org/index.html

Bumblebee let me get Cranvas installed and running.  

However, I still don't think it's working right, because you need to run each app from the command line using something like "optimus firefox" to start the application.  I think that some of the behind the scenes mechanics of cranvas don't know to prefix the calls with optimus.  

Also, I don't think that a free amazon AWS Ubuntu is the best place to test cranvas, since you're not buying access to a graphics card. 

I'm definitely still learning, but I thought this might be helpful. Hopefully this post will be outdated if/when Nvidia support becomes native to Linux.

Michael Lawrence

unread,
Apr 26, 2013, 1:00:05 PM4/26/13
to cra...@googlegroups.com
On Fri, Apr 26, 2013 at 8:15 AM, Gene Leynes <gle...@gmail.com> wrote:
I was going through my bookmarks, and noticed this bookmark from last year and thought I would update.

I got cranvas "working" this week. I'd like to share a little about what I've learned.  Cranvas and it's dependencies use the computer's hardware in a very direct way, especially video card.  So, that has to be in place and and working in the way it expects for things to work.

Since the conference I've purchased a beastly computer with a huge GPU, and I've been anxious to try out Cranvas.... but it still didn't work out of the box in a virtual machine.  It also didn't work when I installed Ubuntu 12.04... or 12.10, or 13....

After a great deal of searching, it turns out that my computer uses something called Optimus which "intelligently" switches between the power hungry Nvidia GPU and the power efficient Intel GPU.  This causes problems with the drivers on Linux, and the default Noveau drivers don't work with the graphics card.  They don't actually recognize the graphics card at all, but it does actually use the less powerful Intel GPU (I think) and doesn't touch the Nvidia card.  Incidentally, the 13.04 does seem to know that it's using the Intel GPU, but it has other problems.

Updating the Nvidia card's drivers using just about any method you can find on the Nvidia website will cause big problems.  Basically the desktop environment gets hosed in one way or another, and you can only do command line stuff.  Sometimes I found ways to repair this, but reinstalling the OS was easier.

For most applications the answer turns out to be something called Bumblebee, which is developed by these really awesome people in the UK.  http://bumblebee-project.org/index.html

Bumblebee let me get Cranvas installed and running.  

However, I still don't think it's working right, because you need to run each app from the command line using something like "optimus firefox" to start the application.  I think that some of the behind the scenes mechanics of cranvas don't know to prefix the calls with optimus.  


Well, I don't know anything about optimus, but cranvas should not be starting other processes, so I would think "optimus R" would suffice. You'd might hit complications though during installation.

Michael
 
Also, I don't think that a free amazon AWS Ubuntu is the best place to test cranvas, since you're not buying access to a graphics card. 

I'm definitely still learning, but I thought this might be helpful. Hopefully this post will be outdated if/when Nvidia support becomes native to Linux.

On Wednesday, August 1, 2012 1:48:53 PM UTC-5, Gene Leynes wrote:
Hello, I was hoping to get help with an ubuntu install.

Since the R conference I've been excited to try this, only to be thwarted by my un-cool Windows OS, and then my paltry 32 bit CPUs (which is so old that it can't even run a 64 bit VM).

Then I realized that I could do it in the cloud!  

[Fast forward two weeks and though 9 Ubuntu instance setup attempts] 

Now I have an Ubuntu instance running in the cloud that I set up only to run cranvas and R Studio, but I still get an error during the installation.

Warning in fun(libname, pkgname) :
  Unable to load library icui18n "Cannot load library icui18n: (libicui18n.so.48: cannot open shared object file: No such file or directory)"
Warning in fun(libname, pkgname) : qtbase: cannot connect to X server
ERROR: loading failed
* removing '/home/ubuntu/R/x86_64-pc-linux-gnu-library/2.15/qtbase'

I've followed the directions on the git wiki, so can someone offer advice?  I've tried a few different things without luck.  Googling icui18n didn't help either.

I'm happy to set up a different instance type (e.g. Linux) for testing.  

Thanks!!

Gene

Instance details:
AMI: ubuntu/images/ebs/ubuntu-precise-12.04-amd64-server-20120616 (ami-82fa58eb)

Installation Messages: 


--
You received this message because you are subscribed to the Google Groups "cranvas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cranvas+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Gene Leynes

unread,
Apr 26, 2013, 1:13:28 PM4/26/13
to cra...@googlegroups.com
Everything seems to be (finally) installed.
I was doing `optimus rstudio` and ` optirun -b primus rstudio`, but I don't know if the qt libraries (and whatever else) are utilizing the graphics card.
I suspect that it's not fully working since it took a long time to load 1e5 rnrom data points, but obviously that's not conclusive evidence.

I was just trying to shed some light on the Optimus thing for future users.  Also, share my conjecture that the amazon instances don't have graphic cards (at least the free ones), so cranvas might not install correctly there.


Tengfei Yin

unread,
Apr 26, 2013, 3:42:54 PM4/26/13
to cra...@googlegroups.com
Thanks for sharing, bumblebee looks interesting. I got an optimus on my thinkpad T510 too, my experience is that never try to install any fancy NVidia driver from PPA or other place, most time, default driver shipped with 12.04, 12.10, 13.04(I have been testing all this before) will always work fine, cranvas and qtpaint work fine too. In my case,additional  nvidia driver always cause severe problem and even break my system.... 

Tengfei


--
You received this message because you are subscribed to the Google Groups "cranvas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cranvas+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 



--
Tengfei Yin
MCDB PhD student
1620 Howe Hall, 2274,
Iowa State University
Ames, IA,50011-2274


Gene Leynes

unread,
Apr 26, 2013, 5:09:15 PM4/26/13
to cra...@googlegroups.com
On Fri, Apr 26, 2013 at 2:42 PM, Tengfei Yin <yinte...@gmail.com> wrote:
Thanks for sharing, bumblebee looks interesting. I got an optimus on my thinkpad T510 too, my experience is that never try to install any fancy NVidia driver from PPA or other place, most time, default driver shipped with 12.04, 12.10, 13.04(I have been testing all this before) will always work fine, cranvas and qtpaint work fine too.

For me the default drivers wouldn't work because the dependencies for cranvas / qtpaint couldn't be satisfied.   I think my graphics card is kind of a new model or something: it's a GE Force 670m.

For you is cranvas fast?  Just out of curiosity, how long does it take you to run this?

library(cranvas)
df <- data.frame(
x = rnorm(1e5),
y = rnorm(1e5),
z = rnorm(1e5),
fac = sample(letters,1e5,replace=TRUE))
system.time(dftest <- qdata(df, color=x, size=y))

For me it's about 13 seconds (which seems slow for a 3GB graphics card).

 
In my case,additional  nvidia driver always cause severe problem and even break my system.... 


Exactly my experience.  Being new to Linux I thought it was user error (me).  After many web searches and trial and error I found that there are a lot of suggestions out there, and most of them are specific to their system and fail for the general case.  The best guide was actually the one that was on the Ubuntu website in the first place: https://help.ubuntu.com/community  After a few clicks, I found myself on a guide that suggested Bumblebee... but this was only after I contacted MSI and Nvidia and learned that I have Optimus, which I didn't know was a problem.

Tengfei Yin

unread,
Apr 26, 2013, 5:22:35 PM4/26/13
to cranvas
Hi 

I see the problem here,  it's slow, because the color scale process based on 1e5 number  is slow, that's a function from scales package, maybe it's being used in the wrong way or maybe it's not optimized, need to dig down, if you remove color = x in your qdata that will be fast, remove size = y will be a little faster,  For R profile, check

$by.total
                          total.time total.pct self.time self.pct
"system.time"                  19.66    100.00      0.00     0.00
"qdata"                        19.54     99.39      0.12     0.61
"cscale"                       18.58     94.51      0.00     0.00
"map_continuous"               18.58     94.51      0.00     0.00
"na_colors"                    18.58     94.51      0.00     0.00
"nice_rgb"                     18.40     93.59      0.00     0.00

cscale consume most time.

Tengfei


--
You received this message because you are subscribed to the Google Groups "cranvas" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cranvas+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 
Reply all
Reply to author
Forward
0 new messages