SciDB R


#1

So I am working with sciDB for my college project. I had a lot of doubts actually that I could not solve from the documentation.

I have installed SciDB as a single node in the VirtualBox with Ubuntu version 14.04 and SciDB version 15.12.
I was trying to get the conversion csv2scidb working which did not work so I downloaded the GitHub io package and placed it in the scidb_src folder. Then I tried the load( output_array, input_file
[,instance_id [,format [,max_errors ,shadow_array [,isStrict ]]]] ) command which did not work. What do I need to do to upload a simple data set to scidb and access it through R?
I should be writing these terminal right? because I can get scidb and R started via terminal of course. Should I be doing this in some different directory other than the default one? (default is scidb@scidb-vm:)

I really need to just upload the data to SciDB and run some simple commands on it via R studio. Can anyone help? I am new to R, Ubuntu and SciDB as well and having a really hard time to understand how to proceed.


#2

Hey - a couple of things

  1. you don’t need to use aio; it mostly helps with strange formats and error tolerance. The load() operator in scidb can take a TSV file as an input

  2. getting plugins to work is pretty straightforward but takes a few steps. A plugin generally compiles to a file named
    lib[NAME].so
    Then that file needs to be copied to
    SCIDB_INSTALL_PATH/lib/scidb/plugins/
    Then you run this command to load it:
    iquery -aq "load_library('NAME')"

There’s also dev_tools - a “plugin that installs other plugins”, available here:


If you read the instructions for dev_tools - you should get a sense of how these things work.

To get SciDB-R to work you will also need shim:

Let us know if you get stuck


#3

So first I download and install Shim.
And as TSV CSV format is supported by SciDB , I don’t need any additional plugins?


#4

If you want to use R, some plugins will enable additional operations: aggregate and join by attribute, faster head(), which is very useful for browsing data, and so on. Once you set up dev_tools, it becomes very easy to add more.

I don’t know how much you will need, generally for 15.12 I do recommend a few common ones.

For a list, see the “Plugins from P4” section of this post:


#5

So coming back to the installation. As I am running SciDB in the virtualBox till now I have been doing the following things:

Installed the virtualBox
In the Terminal I gave the following commands:

  1. $ scidb.py startall mydb
  2. $ sudo service rstudio-server start

And then I opened the rstudio in browser: http://localhost:8787

In the rstudio I put the commands:
install.packages(“scidb”)
library(“scidb”)
scidbconnect()
scidblist()

The output was:
install.packages(“scidb”)
Error in install.packages : Updating loaded packages

data = scidb(“MODIS_DATA”)
Error: could not find function "scidb"
scidbconnect()
Error: could not find function "scidbconnect"
scidblist()
Error: could not find function “scidblist”

What is missing here? Does this mean that sciDB is not configured?

Also, I can access shim via http://localhost:8083. Is that okay? Or do I have to still get shim via the github link you provided?