[Live-demo] ipython upgrade via pip

Brian M Hamlin maplabs at light42.com
Sat May 20 15:30:32 PDT 2017


Jupyter update in OSGeo-Live pre11
20May17 -dbb
====================================

OS Environment:
 OSGeo-Live  (May 10 09:33)  osgeo-live-nightly-PR161-amd64-daace19.iso

--
user at live-pre11:~$ sudo apt-get install libpython2.7-dev
  ... 
user at live-pre11:~$ sudo pip install --upgrade notebook
  (Python 2.7 package change details summarized)

REMOVED:  none
UPDATED:                vers_prev       vers_new
  ipython               4.0.2           5.3.0
  ipykernel             4.2.2           4.6.1
  jinja2                2.8.0           2.9.6
  jupyter-client        4.1.1           5.0.1
  jupyter-core          4.0.6           4.3.0
  notebook              4.2.1           5.0.0
  pyzmq                 15.0.2          16.0.2
  tornado               4.2.1           4.5.1
-
  ipython-genutils      0.1             0.2
  nbconvert             4.1.0           5.1.1
  nbformat              4.0.1           4.3.0
-
  decorator,enum34,jsonschema,Markupsafe,mistune,pexpect
  pickleshare,ptyprocess,Pygments,pyparsing,python-dateutil,traitlets

ADDED:
  appdirs
  backports-abc
  backports.shutil-get-terminal-size
  bleach
  certifi
  configparser
  html5lib
  packaging
  pandocfilters
  pathlib2
  prompt-toolkit
  scandir
  testpath
  wcwidth
  webencodings

 
===============
Notes:

  * the primary change in ipython CLI version is  4.0 -> 5.3
  * jupyter-client  4.1 -> 5.0
  * jupyter-core    4.0 -> 4.3
  * notebook meta-package  4.2 -> 5.0
 
* Many minor utils are also minor updates along stable projects
  in other words, the minor utilities needed for the new setup
  are not breaking changes and unlikely to be breaking changes. 

* No one can predict where in the chain of python interdependancies
  one package may declare strict version requirements instead of
  "greater than or equal to" dependancies, however the python
  ecosystem seeks to stabilize itself generally. This combination
  of package updates tends to work in the modern time. 
  The Debian packaging python chain is not the only definition
  of a working set of packages. 

* numpy, matplotlib, pandas and other major SciPy components,
  were not touched

* within the very active and well-managed IPython and Jupyter
  projects, component version numbers are not entirely harmonized
  to this day. 

--
  some basic testing to follow
  best from Berkeley
   -Brian M Hamlin

 



More information about the Live-demo mailing list