Interesting project. I would've liked to have heard a brief comparison with Excel - what makes Datasette something that one should reach for over that? Is it just the ability to use SQL in-place?
Right now I'd say the big things are:
1. Size of data. Datasette is a great tool for when your data is too big for Excel but not so big that you need a data warehouse - think 100MB to 10GB range
2. Interoperability and automation. Building tools in other languages that interoperate with an Excel spreadsheet - a web crawler that updates it constantly over time for example - is possible but in my experience pretty hard. Writing to a SQLite file is usually much easier in comparison
3. Publishing data. This is a core concept in Datasette: the ability to publish your data online so other people can explore and interact with it.
4. Plugins. Writing plugins for Datasette is pretty easy, and there are over 100 now: https://datasette.io/plugins - so if it doesn't do something you need, you can add that feature independently of the core project
5. SQLite generally. Some projects are a better fit for a database than a spreadsheet, and SQLite is a /really/ good (often underestimated) database.
We took a look at DS several months ago (I believe it popped up here on HN) but I confess that until I read the linked article I hadn’t fully understood that it was a “hammer”, and a good one at that.
At that time we set it aside in favour of AirTable which for our purposes - to relate this to one of your other comments here - is sort of Access with a web interface plus authentication and (frankly not very granular) authorisation.
The main reason was because it seemed to us that Datasette didn’t have first class support for auth - that it was sort of positioned as a community or internal tool, rather than something we could build out a client interface on. I just had a look at the plugins and noted that there is something for auth, although it too looks rather to be for internal tooling. Is that a fair characterisation, or do we have it wrong? Because we love everything else about DS - open source, built on SQL and SQLite, etc. - and would vastly prefer to use it over AirTable.
The most relevant plugin for what you're describing at the moment is probably the Auth0 one: https://datasette.io/plugins/datasette-auth0
Quite a lot of assembly is required at the moment if you want to do interesting custom things with authentication and permissions - the documentation here is the best starting point: https://docs.datasette.io/en/stable/authentication.html
Yes we generally like the AirTable UX, too, and it’s a good fit for client-facing stuff in addition to internal tooling. It is slow, though, and there are no native desktop or mobile clients. Mainly, though: building a business on it comes with all the usual pitfalls of VC-backed SaaS.
It’s very helpful to hear your thoughts on the project’s trajectory. We’ll take another look at DS and the auth0 module, thanks. It looks very promising and we don’t mind a bit of assembly.
Cancelling our subscription today and migrating to Datasette, hell or high water. Glad this happened now and not 2 years down the track.
- it’s open source
- it allows publishing to the web
- it provides handy web-based tooling, like turning your data into a JSON api
- it provides a user-friendly UI but also lets you dig deeper with plain old SQL (pro to some, con to others)
- apparently has an extensible core with a rich plugin ecosystem which allows it to meet a wide range of needs
Datasette isn't there yet but it's a useful ambition for the future.