umbrella VS sequential-uuids

OP here. This project is implemented in Typescript and Svelte, with support from the thi.ng libraries.

This a is a really wide field. Basically its starts with one of the standard algorithms, like and l-system, a voronoi diagram, or simple combination of sin and cos to draw shapes. Have a look at thi-ng[1], just as example for a library that has implemented a lot of this kind of algorithms. The next step in my process is to think about how to destroy this forms as most of them are used a lot in this field and become boring. Using perlin noise as input to arguments of the algorithms is one way to do this. The other part is to make most all of the input variables easily changeable as most of the time the whole process is like writing the program and then spend a lot of time adapt the parameters until you get some interesting output. So in the end I have a small Svelte app, using Svelte only cause the data binding for the inputs so simple, which has tons of sliders that renders an SVG that can be saved in the end.

[1]https://github.com/thi-ng/umbrella

@thi.ng/ksuid https://github.com/thi-ng/umbrella/tree/develop/packages/ksu...

Disclaimer: not a dba so my terms might not be appropriate

I’ve seen uuid4 which replaces the first 4 bytes with a timestamp. It was mentioned to me that this strategy allows postgres to write at the end of the index instead of arbitrarily on disk. I also presume it means it has some decent sorting.

[inspiration](https://github.com/tvondra/sequential-uuids/blob/master/sequ...)

yum install -y git git clone https://github.com/tvondra/sequential-uuids.git ysqlsh -h $(hostname) --echo-all --quiet \ --file sequential-uuids/test/sql/uuids.sql | sdiff sequential-uuids/test/expected/uuids.out -

If you want to keep locality when using uuids, use a different generation function such as: https://github.com/tvondra/sequential-uuids

OP, have a pretty large B2B SAAS app with all UUID pkeys, and the largest issues i'd say with UUID compared to bigint are: size, index bloat, WAL bloat, much slower GIST indexes (for exclusion constraints). You can work around some of the issues (WAL bloat, index bloat) by using a better behaved UUID generation function: https://github.com/tvondra/sequential-uuids You can get something similar for your app side of things if you need to generate UUIDs there.

Use this: https://github.com/tvondra/sequential-uuids

If you cannot install extensions, I just wrote a PL/PGSQL implementation for the time-based generator I could share.

Just fyi, UUID in Postgres is not 40 text chars, it's 16 bytes binary and has a canonical text representation. Also, you don't need to use UUIDv4, I am quite partial to: https://github.com/tvondra/sequential-uuids