Robert Howe
@rchowe.com
Engineer, travel aficionado, and public figure. But I'm really just here to meme.
You should be fine if you're just mapping abc -> JSON.
January 22, 2026 at 8:45 PM
You should be fine if you're just mapping abc -> JSON.
I used it as the backend for pmcal.net (since it was the project I built when I wanted to try Amplify, and the Amplify documentation just pretends it's a normal database) and regret it. At some point I may vibe-code a Postgres replacement for the backend.
January 21, 2026 at 9:52 PM
I used it as the backend for pmcal.net (since it was the project I built when I wanted to try Amplify, and the Amplify documentation just pretends it's a normal database) and regret it. At some point I may vibe-code a Postgres replacement for the backend.
If you are using it for a document-oriented data store, it's not bad. If you are using it as the backend for tabular data, it's pretty bad.
January 21, 2026 at 9:52 PM
If you are using it for a document-oriented data store, it's not bad. If you are using it as the backend for tabular data, it's pretty bad.
There are a lot of footguns in DynamoDB...
1. A 1 MB / 1000 records max response size.
2. Limit refers to the number of records scanned, not the number of records returned.
3. There are pretty much only two read operations: forward table scan and index lookup.
1. A 1 MB / 1000 records max response size.
2. Limit refers to the number of records scanned, not the number of records returned.
3. There are pretty much only two read operations: forward table scan and index lookup.
January 21, 2026 at 9:52 PM
There are a lot of footguns in DynamoDB...
1. A 1 MB / 1000 records max response size.
2. Limit refers to the number of records scanned, not the number of records returned.
3. There are pretty much only two read operations: forward table scan and index lookup.
1. A 1 MB / 1000 records max response size.
2. Limit refers to the number of records scanned, not the number of records returned.
3. There are pretty much only two read operations: forward table scan and index lookup.
Can confirm, GIL sucks.
We accepted the restrictions that come with Python multiprocessing and MPS to get around it.
We accepted the restrictions that come with Python multiprocessing and MPS to get around it.
December 18, 2025 at 5:41 PM
Can confirm, GIL sucks.
We accepted the restrictions that come with Python multiprocessing and MPS to get around it.
We accepted the restrictions that come with Python multiprocessing and MPS to get around it.
Five months later, 27 JetBlue flights, 6 non-JetBlue flights, and three trains. I just touched down in Manchester NH as JetBlue city 25.
This is what the map ended up looking like. I am pleased to say that I never missed a connection or was significantly affected by a delay.
This is what the map ended up looking like. I am pleased to say that I never missed a connection or was significantly affected by a delay.
December 12, 2025 at 4:59 AM
Five months later, 27 JetBlue flights, 6 non-JetBlue flights, and three trains. I just touched down in Manchester NH as JetBlue city 25.
This is what the map ended up looking like. I am pleased to say that I never missed a connection or was significantly affected by a delay.
This is what the map ended up looking like. I am pleased to say that I never missed a connection or was significantly affected by a delay.
Also in charge stats: I spent $2,169 on supercharging this year, which I feel is not a bad fuel bill since I drive a ton and almost exclusively charge at superchargers.
December 11, 2025 at 2:17 PM
Also in charge stats: I spent $2,169 on supercharging this year, which I feel is not a bad fuel bill since I drive a ton and almost exclusively charge at superchargers.
Don’t forget the MTA Staten Island Railway
December 9, 2025 at 9:46 PM
Don’t forget the MTA Staten Island Railway