@Jonathan_Blow 100% agree. Not a universe, yet and maybe never, where AI generated code can be in our codebase. AI is useful solely as a basic research assistant (show me the return shape of xyz api) and example generator (let me see the syntax for xyz that we dont use often again). That is all
@WarrenInTheBuff Timing matters. 99% of the time, value, if created at all for start-ups, will only occur at sufficiently low exercise prices. If you aren't in early, options usually aren't great. RSUs only happen early, if at all, as they are taxable events. Lot's of other factors to consider.
@htmx_org Once again, the old school ssr mindset is the right one. Ended up not using htmx, but this mindset was followed, and Vue is just an advanced template renderer, as it should be. ~200 api endpoints and no front state / store because it isn't needed if done right. That paper to come
@Thomasmarkelly@htmx_org We've found that granularity and isolation are key. Isolated complexity when composed across multiple apps / models is very powerful. Models broken out into whatever logical steps your app has where SELECT EXISTS (boolean state) becomes the easily obtained truth scales.
@htmx_org We've found that granularity and isolation are key. Isolated complexity when composed across multiple apps / models is very powerful. Models broken out into whatever logical steps your app has where SELECT EXISTS (boolean state) becomes the easily obtained truth scales.
@htmx_org Ultimately, even if you don't use htmx, the mental model it reinforces, "backend-for-frontend," will make any app better, faster, and more maintainable. This model forces complexity to be moved where it belongs: the backend and db. Devs should try to build something hard with it!
@RyanCarniato @all__hype You obviously know a lot more about this than I do, so thanks for the detailed background and overview. In the end, as a consumer of these tools, again, I am very glad to see that a positive shift has occurred and excited to see it continue.
@RyanCarniato @all__hype Very fair assessment. There was a lot of uncertainty around the "best" approach to building apps at the time. Maybe a bit of browser performance mixed with a specific dev ergonomics goal? Whatever the reason, very glad to see a "solid" shift to a cleaner approach.
@all__hype Running for 100 years is vastly different than evolving and expanding for 100 years. This is where he is incredibly wrong, in my opinion, unless he means "managing" to mean maintainence but I don't get that vibe.
@htmx_org I'll go out on a limb here and say that you cannot build any app without coupling. The choice is where the coupling occurs. It's always the young React folks who think they know everything.
@AutisticOvrflow@teej_dv There is still a thing called the backend owns state, even with a frontend backend split. This is pure insanity and would honestly fire the developer immediately. This is not how you build an app that will ever scale properly.
@htmx_org If you are writing apps where the current state of LLMs can write what you deem as business differentiating logic / code for you, then strongly reconsider the app you are building. And yes, that's harsh, but assure you it is not an uninformed comment.
@htmx_org @lurkingforev @SMT_Solvers Good orm's are a serious value. Laravel, django, ruby on rails, etc. The final developed queries are super efficient if implemented the right way. In my experience, it's easier to create an optional query with a good orm than with raw sql and easier to maintain. The latter is key
One of the most enjoyable software dev moments is when you battle with something for an hour, days, weeks, and, finally, a pattern reveals itself, and what felt impossible suddenly feels so simple and crystal clear. It's one of the biggest and rare joys that keeps me coding.