WuPlus
commented
9 years ago -------From Evgeny Karataev Users of our system should not know about our back end. E.g. if are are running one, two or more servers. So they should not be expected to "switch" between two systems.
"We can use the web UI that IPython provides.(We can customize it )" - yes, we need to reuse their UI, but it should be integrated with our UI (e.g. a list of available data sources). Also the "customization" should be done carefully, so we can easily update to newer versions once they release a new version. Our UI can be simple. A list of available data sources where user can click and get more info about a datasource, then click another button to open IPython notebook with that selected data source being available in the IPython for the analysis.
"If IPython need the data, it can ask our Jetty Server via RESTful API" - I would suggest to start with this. Create a test local database and try to feed it to IPython. Later when we have our MetaStore project running, the list of available data sources will come from them via RESTful request.
"RESTful APIs also provide notebook manipulation functions." - as I remember after last time I played with the IPython Notebooks, it already has notebook management functionality, e.g. you can see a list of notebooks you created before, create a new or delete existing one. We should probably just use that, but there is one thing here. Out system will be multi user system and one user should not see IPython Notebooks that belongs to another user. Do u know what I mean?
Let me know if you have any questions. Try to write a general description of the project and post it GitHub in the readme.md file.
Here is my rough solution to implement our IPython project.