Not Theo

[Not Theo @hottakeweekly]

@ggs4ns That's not fair!

[Not Theo @hottakeweekly]

Alright who is using agents?

[Not Theo @hottakeweekly]

@g2sans @theo @t3dotchat Violation: Measuring performance and creating an educational resource to back up your claims.

This is not allowed in frontend discourse.

[Not Theo @hottakeweekly]

@g2sans @theo @t3dotchat 🚨STOP!

[Not Theo @hottakeweekly]

@dmitriid @AdamRackis Most of the grief is with the complexity introduced in modern frameworks most of us have been working with for the last 10 years across many teams/approaches and day-to-day experience using apps built this way that don't work reliably.

[Not Theo @hottakeweekly]

@dmitriid @AdamRackis fwiw I agree with the article. ime most modern frontends aren't built practicing hypermedia in practice. Most of my Datastar apps continue to work even if Datastar fails to load.

[Not Theo @hottakeweekly]

@asilvadesigns @DelaneyGillilan Wait until I get around to optimizing it!

[Not Theo @hottakeweekly]

@DelaneyGillilan You can't do that with Datastar!

[Not Theo @hottakeweekly]

@grok @CFchangelog @grok can you enumerate all the CVEs related to modern frontend frameworks from the last year?

[Not Theo @hottakeweekly]

@AdamRackis @DelaneyGillilan I promise, I tried logging out of x, but they use optimistic updates for that and it never actually works.

[Not Theo @hottakeweekly]

@AdamRackis Like why do they get so upset at stuff like this? https://t.co/fdFkUmDUTi

[Grok @grok]

Based on the Cloudflare WAF note about active RCE via invalid path sequences in popular frameworks, plus the same-day Next.js/React security patches for critical unauth RCE in Cache Components and RSC paths, my bet is RSC (affecting Next.js apps). Vercel did ship the update. Patch ASAP if affected.

[Not Theo @hottakeweekly]

@AdamRackis Got 'em!

[Not Theo @hottakeweekly]

@dhh Sir this is a Wendy's

[Not Theo @hottakeweekly]

@xmlking @g2sans @tannerlinsley @sveltejs @tan_stack It's been a minute since I wired it up and have learned a couple thing though. I think that's generally the right way to do it (let server know what cache keys are live and push ops to those). Oh no you gave me ideas.

[Not Theo @hottakeweekly]

@xmlking @g2sans @tannerlinsley @sveltejs @tan_stack Yeah I am using tanstack-db. I guess I just don't like the model of pushing invalidation to then re-query, but it definitely is a workable approach. https://t.co/Uv0f63Ne9w is an interesting exploration in this.

[Not Theo @hottakeweekly]

@HugeLeters @g2sans @superhighfives @nicu_tsx @tannerlinsley data-* is just a string expression that hooks in to an engine (Datastar) and is executed on demand.
if I can encode my behavioral intent those strings just ride same same html patch I am sending compressed (doesn't really add bloat). the amount of JS in any page remains bounded.

[Not Theo @hottakeweekly]

@HugeLeters @g2sans @superhighfives @nicu_tsx @tannerlinsley If talking about rendering components on the client:
- Every feature adds more JS to my overall bundle size
- I have to have a contract for sending JSON to the client
- React has to consume that data, do a vdom diff then dom operations

[Not Theo @hottakeweekly]

@RyanCarniato @devagrawal09 @AdamRackis I would measure the specific use case though. If re-rendering the whole page does become a bottleneck I adjust and optimize. Maybe I am only morphing parts of the UI that are necessary. I don't have to default to that though.

[Not Theo @hottakeweekly]

@RyanCarniato @devagrawal09 @AdamRackis I can even just push an update every tick. I'm doing cheap db calls and concatenating strings. Not very taxing on servers we're deploying to these days.