4 ms·
Nice to see another player in the crowded hosted Jupyter notebooks space! Still very far from smooth (<10) click experience. The workflow took a while to get
by sireat 6y ago
Nice to see another player in the crowded hosted Jupyter notebooks space!
Still very far from smooth (<10) click experience.
The workflow took a while to get started:
Manually create account (no external auth option such as Github?) -> verify account -> create notebook -> set pw for notebook -> run notebook
I did all that only to find that no required libraries are there.... not even Pandas?
By comparision Google Colab onboarding is much smoother and you know you are getting a working Tensorflow, Pandas, Plotly etc out the box. Sure you pay the privacy tax to Google for this smoothness.
My biases:
- I've installed and maintain Littlest Jupyter Hub instance for teammates at my organization.
- I teach using local Anaconda installation.
- I use myBinder and Google Colab.
If I did not have a smooth experience I would imagine someone starting out would have a harder time.
At the very least you should curate the packages so they run out of the box, without fiddling with requirements.txt, yaml, pip install, conda etc.
- sanketsarang 6y agoThis is good feedback. Thank you for documenting your experience. Basis this we concluded on the following. Adding features: - Login with Google & Github accounts - Changing default type setting to "Data Science Notebook" - A package library with one-click install Changing features: - Simpler UX (reduced clicks required) - Removing the mandatory plan selection - Starting every user on free plan by default Anything else you recommend we should do?
- sanketsarang 6y agoSince you are an active user of myBinder and Colab, I am tempted to ask. What are those features we can add that will make you let go of myBinder and Colab?