What do you like best about Datalore?
-Great type-hinting and code completion you would expect from a JetBrains IDE.
-Great integration of external code repositories, with easy management of SSH keys. This has enabled us to accelerate our data processing by easy access to git managed tooling within Datalore.
-Reports and widgets generally within notebooks are great tools for creating simple scripts for a non-coder to execute to analyse data and produce results when desired. This has enabled us to free up data scientist resources for simple requests to reprocess similar data over and over.
-Cloud storage integration mounts storage onto the notebook making accessing data as simple as accessing local data is!
-Tech support is extremely quick to resolve a problem. Recently I opened a ticket about a specific install issue relating to a custom repo we had made. Within a couple days the tech support team had identified the issue with Datalore (conflicting Python versions) and rolled out a patch to fix the issue. Cannot beat the speed and effectiveness of their tech support - only praise!
-Billing is easy to keep track of, especially with itemised cost and storage usage that can be exported as a .csv for analysis (with a Datalore notebook itself).
-Database connection is as easy to setup as cloud storage. SQL cells are easy to use. We did find they ran quite slow, making them not useful for accessing larger databases and resulted to 'pandas' access manually. However I believe a patch sped these up, and I believe it was significant - though we have only utilised it in a limit capacity since the update.
-Notebook sheets and table of contents in particular, allow for easy organisation of notebooks via markdown headers that show on the ToC - making navigation of larger notebooks easy.
-Collaboration is great due to web-hosting of notebooks, allowing multiple users to edit a notebook simultaneously without any conflicts. Review collected by and hosted on G2.com.
What do you dislike about Datalore?
-Python version is at 3.8 on notebooks. Have wanted to see a more modern version or options for a while, and despite plans to support multiple version from Datalore for over a year this hasn't happened.
-There are often issues accessing spot instances regularly for the CPU-XL tier forcing usage of on-demand or alternative computes, making spot CPU-XL almost unusable for us as an option.
-Minor gripe: markdown has spellchecking but cannot right click to fix the typo.
-On a wishlist: would like reports to be able to rerun a subsection of a notebook when called to (maybe a button widget), rather than having to rerun the entire notebook: this makes interactive reports unusable when having to load or process large data (since this is only needed once but reruns every small change). Review collected by and hosted on G2.com.
