IPython dependency in the install script
Hi all, With the release of Jupyter, the number of dependencies necessary for the IPython notebook has exploded somewhat. I just created a new virtualenv and did "pip install jupyter". The output of "pip freeze" for this virtualenv is pasted here: http://paste.yt-project.org/show/6141/ As you can see, jupyter brings in a large number of python packages, and it would be a real pain to manually install all of these in the install script. This is coming up because I'm in the middle of updating the notebooks in our docs to use the latest notebook format. To make it so users can *read* these notebooks, I'd also like to update the version of IPython/Jupyter being installed. I can see three ways forward, in order of my preference: 1. Use pip to install jupyter, which will automatically bring in IPython and all its dependencies. I'd probably pin it at Jupyter 1.x / IPython 4.x to avoid having it update to a possible incompatible notebook version in the future. 2. Manually install Jupyter's dependencies in the install script, as we've done for IPython in the past. 3. Punt and just keep on installing IPython 2.4.1. What do people think about this? Would anyone have objections to relying on pip to install Jupyter and its dependencies? Thanks for your advice, -Nathan _______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
My preference is for option 1. Option 3, in my view, should not even be an option. The install script is just going to get more difficult to maintain as time goes on…
On Jan 12, 2016, at 5:44 PM, Nathan Goldbaum
wrote: Hi all,
With the release of Jupyter, the number of dependencies necessary for the IPython notebook has exploded somewhat. I just created a new virtualenv and did "pip install jupyter". The output of "pip freeze" for this virtualenv is pasted here:
http://paste.yt-project.org/show/6141/ http://paste.yt-project.org/show/6141/
As you can see, jupyter brings in a large number of python packages, and it would be a real pain to manually install all of these in the install script.
This is coming up because I'm in the middle of updating the notebooks in our docs to use the latest notebook format. To make it so users can *read* these notebooks, I'd also like to update the version of IPython/Jupyter being installed.
I can see three ways forward, in order of my preference:
1. Use pip to install jupyter, which will automatically bring in IPython and all its dependencies. I'd probably pin it at Jupyter 1.x / IPython 4.x to avoid having it update to a possible incompatible notebook version in the future.
2. Manually install Jupyter's dependencies in the install script, as we've done for IPython in the past.
3. Punt and just keep on installing IPython 2.4.1.
What do people think about this? Would anyone have objections to relying on pip to install Jupyter and its dependencies?
Thanks for your advice,
-Nathan _______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
I've opened a pull request for the notebook updates here:
https://bitbucket.org/yt_analysis/yt/pull-requests/1935/wip-update-documenta...
I'll wait until we come to a consensus in this thread before removing the
WIP tag.
On Tue, Jan 12, 2016 at 4:48 PM, John ZuHone
My preference is for option 1. Option 3, in my view, should not even be an option.
The install script is just going to get more difficult to maintain as time goes on…
On Jan 12, 2016, at 5:44 PM, Nathan Goldbaum
wrote: Hi all,
With the release of Jupyter, the number of dependencies necessary for the IPython notebook has exploded somewhat. I just created a new virtualenv and did "pip install jupyter". The output of "pip freeze" for this virtualenv is pasted here:
http://paste.yt-project.org/show/6141/
As you can see, jupyter brings in a large number of python packages, and it would be a real pain to manually install all of these in the install script.
This is coming up because I'm in the middle of updating the notebooks in our docs to use the latest notebook format. To make it so users can *read* these notebooks, I'd also like to update the version of IPython/Jupyter being installed.
I can see three ways forward, in order of my preference:
1. Use pip to install jupyter, which will automatically bring in IPython and all its dependencies. I'd probably pin it at Jupyter 1.x / IPython 4.x to avoid having it update to a possible incompatible notebook version in the future.
2. Manually install Jupyter's dependencies in the install script, as we've done for IPython in the past.
3. Punt and just keep on installing IPython 2.4.1.
What do people think about this? Would anyone have objections to relying on pip to install Jupyter and its dependencies?
Thanks for your advice,
-Nathan _______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
Hi Nathan,
The age of micropackages is upon us!
Since we don't depend on Jupyter/IPython as a hard dependency, I think
we should go ahead and use pip.
-Matt
Relevant: https://twitter.com/twiecki/status/687029641333354497
On Tue, Jan 12, 2016 at 4:44 PM, Nathan Goldbaum
Hi all,
With the release of Jupyter, the number of dependencies necessary for the IPython notebook has exploded somewhat. I just created a new virtualenv and did "pip install jupyter". The output of "pip freeze" for this virtualenv is pasted here:
http://paste.yt-project.org/show/6141/
As you can see, jupyter brings in a large number of python packages, and it would be a real pain to manually install all of these in the install script.
This is coming up because I'm in the middle of updating the notebooks in our docs to use the latest notebook format. To make it so users can *read* these notebooks, I'd also like to update the version of IPython/Jupyter being installed.
I can see three ways forward, in order of my preference:
1. Use pip to install jupyter, which will automatically bring in IPython and all its dependencies. I'd probably pin it at Jupyter 1.x / IPython 4.x to avoid having it update to a possible incompatible notebook version in the future.
2. Manually install Jupyter's dependencies in the install script, as we've done for IPython in the past.
3. Punt and just keep on installing IPython 2.4.1.
What do people think about this? Would anyone have objections to relying on pip to install Jupyter and its dependencies?
Thanks for your advice,
-Nathan
_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
_______________________________________________ yt-dev mailing list yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
participants (3)
-
John ZuHone
-
Matthew Turk
-
Nathan Goldbaum