Undb – Private first, unified, self-hosted no code database
github.com
github.com
NocoDB is neat, but looks like it's a single vendor-driven project that could go open core or die altogether if funding dries up.
Baserow also looks good, but already appears to be open core and another single-vendor project.
Undb might be a more community driven project over time. Hard to say - it looks like there's 6 contributors currently and there's not much info I could find by way of governance or company info to determine where they want to go.
Single-vendor open core (SVOC) can move quickly but it tends to degrade and disappoint over time. I would love to see a project of this sort that's actually trying to be truly community driven.
That's probably unrealistic, but I don't know if there's a ton of value in an open core solution over just using Airtable or Notion if some of the best bits are closed anyway and the fully closed alternatives are more polished and straightforward.
That may be true. Or it may stall development of a truly open project, if the open core is widely adopted and people accept it as good enough until what's considered "core" shrinks (e.g. features closed that used to be open) or some other shift.
I'm philosophically opposed to open core (Mathesar is run out of a non-profit), but I can see why other projects do it – finding funding for a big project like this is difficult without VCs (who expect returns).
I've been in the midst of developing my own frontend (SPA/PWA) data storage solution, using IndexedDB (and the Dexie library), so I've been watching all of the recent announcements about the wasm-sqlite implementations and wondering if I should probably just switch over to that. If, for nothing else, for the exportable .sqlite file that would be available without having to do a custom "Database Export".
With that in mind, I'd be much more interested in a clean library that abstracts sqlite data access, but stores it in the same format. That way, I could implement the library in my frontend app to store data, and then use the sqlite files that can be exported to load into your Undb app to inspect them (I'm thinking for, say, error debugging or whatever). Similar to how I might pull a client's db backup and load it into Microsoft's SSMS management app.
But hey, maybe that's in the roadmap! And, even if not, it's still a great utility all on its own. And it seems to be a solidly made and well-styled app. Very interested to see if other people have some obvious use cases for this.
That said, in comparison to other similar software:
- For standalone use: there's a lot of similar software like budibase, baserow, nocodb, etc. which is heavier weight (distributed as a docker container) but more featureful
- For use with a custom frontend: I'm not sure I would pick it when it's AGPL, since PocketBase is roughly comparable as a single-file sqlite-based backend and it's MIT license.
Some ui-breakpoints for a mobile view on tables are missing and not yet catalogued in the issue tracker. Use-case: string searching a table on the go.
I thought about some very trivial apps that are worth building but not worth the cost of the usual stacks. We are talking about the equivalent of a simple desktop app that doesn't need hosting except you don't want to use the local filesystem because you want to synchronize your data between devices. The actual app could be a simple SPA. It doesn't matter how it is implemented. What I am missing is the storage concept. There is remoteStorage.js but nobody adopted it.
We have Sheets for data ingestion, and other sheets that do the business logic.
This turned out to be a surprisingly well working pipeline and comes at zero hosting costs. We could just run Jupyter Lab somewhere, but this would cost too much and not nearly as quick to handle for our case.
I currently use NocoDB with a MariaDB database, so I can't easily switch over as Undb doesn't support MySQL as I see. But it has the feature I'm missing in NocoDB: a Calendar View which funnily NocoDB promoted on their website but doesn't have [1]. I guess I'll give it a try!
We just managed to handle all date time and timezone issues which is quite complicated and made a release like a month ago. When we started, we could not have anticipated the difficulties surrounding timezone issues. So we will be getting to Calendar view soon.