re-frame

2026-06-17T02:06:52.590429Z

@rolthiolliere dereferenced subscriptions inside the signal function is too much of break with re-frame2 ethos, but, having said that, I think the version I susggested will work for you. Subscriptions are just a functional view of db and, if you have db , you can apply the functions.

2026-06-17T02:08:08.341849Z

Just to be clear that's not there now, but I could make that change.

2026-06-17T02:10:37.138879Z

BUT, to be honest, my Spidey senses are tingling a bit with this kind of code. It seems like this should almost be recorded as state in app-db, and put there via flows? Maybe.

2026-06-17T02:11:59.176609Z

In terms of materialised views of application state, flows are kinda like subscriptions, except the result is recorded in app-db

rolt 2026-06-17T08:21:55.938319Z

Maybe flows could work, I'll try to test that. The problem with using the db value as you suggested is that sub would recompute at every db change.

2026-06-17T14:44:46.102579Z

@alidcastano I'm acutely aware that I've lost control over the code. It feels odd. I've currently got no idea how various features of re-frame2 are coded. I know what the design is, and I've made literally a 1000 design choices, but I have no idea about the code. So, I'm "trapped" using AI to maintain it. But equally, I have no idea how the compiler creates bytecode either. I couldn't edit that bytecode myself. So, I see it as layers. The engineering effort has moved up a layer. I used to be an expert in 8086 assembler about 40 years ago too (near and far pointers), but value moved up the chain. That was a hard expertise to give up too.

☑️ 1