James Arthur

I guess one aspect here is whether PGlite loads the whole database (indexes and all) into memory. Which I believe it does right now? As opposed to somehow keeping data...

@3dyuval absolutely, that would be awesome 🔥

Note that if we're changing it then we'll need to update a few deployment templates / guides as well.

N.b.: https://discord.com/channels/933657521581858818/1285476835412541516

As per my previous discord note about diffing, the payload format of: ``` data: {"value":{"id":"9a33ad9e-39a6-4ab4-aebf-4a4c54c8dbc6"},"key":"\"public\".\"items\"/\"9a33ad9e-39a6-4ab4-aebf-4a4c54c8dbc6\"","headers":{"last":true,"relation":["public","items"],"operation":"insert","lsn":"5501319691220014840","op_position":0,"txids":[792]}} data: {"headers":{"control":"up-to-date","global_last_seen_lsn":"5501319691220014840"}} ``` To send the smallest insert is not exactly optimal. Perhaps it gzips...

Got it, thanks! Still though, can the Caddy-we-put-in-docker not use the same root CA that caddy trust installs? Then the workflow is “caddy trust once then set env var ongoing”...

N.b.: just eyeballing it in the terminal: ``` ➜ ~ curl -v http://localhost:3000/v1/shape/items\?offset\=26760640_0\&shape_id\=56264740-1729877730504\&live\=true * Trying 127.0.0.1:3000... * Connected to localhost (127.0.0.1) port 3000 (#0) > GET /v1/shape/items?offset=26760640_0&shape_id=56264740-1729877730504&live=true HTTP/1.1 > Host:...

I think this can be solved in documentation. I.e.: use a terse name like `electric-offset` and `electric-cursor` in the headers and then explain in the documentation (we have the openapi...

Yup. Much PRs.

> why can't the offset just be the current Postgres txid? Would something like that allow for load-balancing without stickiness and failover without requiring resync? Right now we query the...