Baserow

TL;DR - It's pretty great.

I was looking for a solution for a publicly facing table to replace a GSheet. The issue being the sheet we needed to display had a changing amount of that was anywhere between 20-60 thousand rows.

While for the intended purpose, an embedded GSheets worked quite well. However, it was extremely inefficient and the load times were unfavourable. As the intent was to display a library of vinyl for ordering / requesting.

The problem being the webform would load, the page would lag and stutter until the GSheet had loaded. So, you've seen the webform and the instructions, but the list of vinyl is nowhere to be seen. Which is less than ideal if you're trying to capture a potential customers expression of interest.

I'm an avid fan of self-hosting, I would argue that I haven't been doing it for the longest time, nor do I have the most complex setup. Primarily, I use Unraid, I feel like I'd get more out of moving to a traditional Linux setup like Ubuntu, chuck Portainer on and away we go. But... I'm a person of convenience, and I'm pretty fond of how Unraid handles docker containers.

So, I was perusing through the Community Applications looking for something that handle tables. There wasn't a lot that stood out mostly your standard databases, ie; PostGres, MongoDB, MariaDB, etc.

Then I saw Baserow, which at a glance just looks like a stereotypical project management service. Nothing to lose, so why not give it a shot.

Initially, I did have issues setting it up behind Nginx Proxy Manager, and if anyone sees this post, these are the settings I forgot on my first attempt -

The frontend would load, but it would have issues connecting to the database. The above resolved my issue. It should have been obvious as I've ran into similar issues before, but I did fumble on this one for a bit.

For both work and home use, I use n8n for a lot of things. I'm not a developer by any means, but it definitely helps for smaller automation tasks or for someone like myself... A lot. But the no code, or I should say little to no code is extremely helpful for almost everything I do. Coupled with some GPT for some code functions, I'll learn one day, but like I said... Convenience.

There definitely is some downsides to using a Zapier-like tool, the predefined functions can be limiting at times. I found that the Baserow integration while very convenient, was very slow. As it would only provide single line requests.

As such, adding and removing 30 thousand rows, even with everything setup locally, would take about 2 hours to remove all the rows, then an additional 2 hours to re-add the rows. Which is how I'm currently going about updating the vinyl list. Reason being if the fresh list from the supplier shows they don't currently have stock, we don't want to display it. It makes for a confusing experience, especially when they're being imported and the order times are quite extensive.

4 hours is a pretty unrealistic down(ish) time, figured I'd look at the API documentation instead see if there's batch remove / add.

And shiiiiiiiiiit, I like how they've done their documentation a lot. It's specific to your instance, the cURL commands are catered to your instance, base URL, database, table, everything. In hindsight, it's probably such a simple thing developers could utilise in their documentation, but this is implemented so well, I can't commend it enough.

This vastly improved my down time. There is a limitation with using the batch mode API calls, it's limited to 200 objects. I find that kind of odd for a self-hosted instance, especially if I have the processing power to easily handle significantly more.

The difference... 2 hours to remove every row, down to 30 seconds. Adding rows wasn't quite as impressive, nevertheless a vast improvement - 2 hours to add, down to 12 minutes.

Scheduling this to run in the AM, I think would inconvenience very few people, and realistically, it won't be blank, it'll just be re-populating.

Anyway, I like collecting physical media, and one of my more unique ongoing collections is Bandai SFC games. For some reason they decided to number these games, and when I collecting more retro Dragon Ball Z games, I noticed the numbering and started looking into it, which has thrown me down a rabbit hole of wanting to own the entire collection.

So, here's my table of all the Bandai SFC games, and how my collection is going.

Moral of the story, GSheets is a perfectly acceptable stopgap, and I don't think anyone should be ashamed. Unless you find some cool ass tables that are easy to use.

Also, why is Cyborg 009 $5,680?! That's absurd.