@Savlambda The issue is always when rubber meets the road and you need to review your architecture and assumptions.
Working from the domain model without insight into the implementation is bound to miss important constraints and edge cases, no?
@devinjameson How are you using it, Devin?
Been eyeing it since Harry was first sharing experiments with Jev-based linting, but still haven't had the clarity of how to use it
@tantaman Idk man, looking at the monumental effort of @tursodatabase makes me think the issue isn't perf. but correctness. How are you balancing this in rindle?
@anselm_io@tursodatabase@EffectTS_ Sorry, was trying to keep the tweet under the character limit 😅
I mean 2-3 months of unencrypted data while y'all ship e2ee. Given your planned timeline, I'll experiment with Jazz already, then!
@anselm_io@tursodatabase@EffectTS_ Of course I'd love it all to be under a single system. Jazz V2's approach seems to fit the bill, but the experiment I made ~3 months ago was lacking so I opted for the more stable SQLite.
Thinking of keeping private data in Jazz with on-device backups as an initial bet.
@anselm_io Will there be an incremental path to adopt it?
In my app, public data is in @tursodatabase / sqlite under @EffectTS_ APIs. Boring CRUD
But private data needs to be e2ee & user-owned. I can take 2-3 unencrypted months. Is adopting jazz advised?
@elianiva_@devinjameson I've already migrated all of mine, it feels pretty stable! Of course, it does imply extra tokens every time a new preview release comes out, but I think it's worth it :)
Anyone doing i18n with @foldkit ?
If we could access Effect services from views, it'd be a trivial problem, but given they need to be pure the story is a bit more complicated.... I'd like to avoid passing in parent context into all child blocks (necessary for model-based locale)
@NathanFlurry Seems like you need to get into @alchemy_run and tap into the new Domain package - being able to deploy only to DO is a disservice to the great ideas there
I also expect Alchemy to be a great source of marketing, it'll grow quite significantly in the next months.
@danieljvdm Ah, and I'm using yielded/auth
By the way, what's your recommendation for local-first auth, or at least sessions that I can keep using offline for later syncing?
There's a write path that is synced, but that I need to be encrypted and so am not considering yielded/sync :)
@danieljvdm I have a use-case for syncing read-only data from a wiki based on users' preferences for what article type to sync.
I've my own simple sync: a cacheableArticleKinds + lastSynced timestamp, then storing what changed since last sync in client SQLite.
Worth migrating over yielded?
@susumartari Parce, dejen de culparle al candidato, pura intervención imperialista. Y no idealizen a Brasil, fuimos dormidos como ustedes, mira nuestro Congresso y Senado!
Lograron elegir a Petro - acuerdo mi asombro con el Paro y la mobilización que tuvieron, pueden volver a hacerlo compas!
@sheherenow_ I'm trying to live this life of providing rural communities with software grounded in their reality 🌱
Almost done with the new portfolio https://t.co/hZwsI8SQb3
Any wisdom jem on communal software?
@devinjameson Thanks for the generous communication, Devin. I'm astonished by your energy 🥹
Is there a case where a Block will have multiple Compositions? If not, can we merge Definition & Composition?
A piece is always defined by how it composes with the surrounding ecosystem, after all :)
@devinjameson One question that I haven't been able to answer myself - not sure if due to LLM-generated docs or my ignorance:
I get why we should split the block's definition from the implementation (insp. Effect Services)... But why split in 3 - composition, preparation and assembly?
@devinjameson Moving `.makeUpdate` from callback + Match.value to a record with a callback per message tag feels so obvious in retrospect!
Still haven't had time to dive deep into this release, have just been pointing my agent at it while I work in the fields, but gut feeling is 🤌