Svelte with Rich Harris

[@paterfamilias01]

ive been studying html, css/scss and vanilla js for the past year. Cant wait to master svelte.

[@ba8e]

Rich Harris... You are a fucking GOD! Thank you for saving my dev life.

[@europeantechie]

really good content, thanks!

[@alexandrodisla6285]

Can listen to him for days?

[@galaxiegalaxie5003]

Great podcast, since i am new to svelte, is there any architecture pattern like mvvm where your viewmodel or store maintain your app state and repository will make an http api call.What is the best pattern to use, where can i look into it.

[@zb2747]

@@internet4543 lmao

[@ricardomuzio2379]

Al día de hoy no hay un ejemplo claro de como conectarse a base de datos como mysql, mariadb o postgresql, como hacer un deploy en vercel o aws, y que funcione correctamente. Habrá alguna documentación con un un ejemplo completo?

[@cognominal]

What is the No Javascript toolkit he cites ? Ashtray ?!?!

[@T--T]

I believe it is Astro

[@CTimmerman]

How is Svelte better than React?

[@yuryzhuravlev2312]

Significant!

[@werethehausers904]

In every way 😉

[@CTimmerman]

So just hype without substance.

[@ssw4m]

Try doing the Svelte online tutorial, and you will find out.

[@IanBuchanan0]

@@CTimmerman Things that are super complicated to do in React are so much easier and simpler to do in Svelte, in my experience.

[@benacker8525]

PodRocket. You straight up asked question like journalist and Rich was speaking 99.9% of the time. Not a podcast at all.

[@irlshrek]

What..?

[@benacker8525]

@@irlshrek this is an interview....

[@irlshrek]

@@benacker8525 my confusion was regarding whether theyre mutually exclusive

[@jeremycoleman3282]

There is a bit of information here. You say svelte does granular updates, so it has less overhead than diffing. But, how does svelte calculate its granular updates? Is there actually some computation cost that was never mentioned? When using granular reactions, does a state change trigger a look-up to which elements rely on this state? Is this memoized (thus diffed)? When you have a large app, is it not more efficient to do event delegation (aka synthetic events) rather than granular watching? Assuming that is true, because it is, what size applications would benefit from svelte compared to react or inferno? Such a metric should likely be in terms of either total elements or event listeners.

[@W4nn3]

The compiled svelte code is very readable, so you can learn a lot about how svelte does things by going to the svelte repl and look at the output. Every Svelte component has an internal variable "ctx" which is an array of all your used variables. For example a variable defined like "let counter = 0;" will internally be referenced with something like "ctx[3]". Another internal variable is "dirty" which is a bitmap or bit array that saves which variables got changed. For example, if you change the variable with "counter += 1;" the compiler transforms this into something like "$$invalidate(8, ctx[3] += 1);" (where 8 is the bit for ctx[3], 4 would be for ctx[2], 16 would be for ctx[4], etc.), the "$$invalidate" function updates the "dirty" variable (something like "dirty |= 8"). In the update cycle, the part that uses the "counter" variable only runs if "dirty & 8". I hope that was somewhat understandable.

[@jeremycoleman3282]

Sorry, my questions were rhetorical and meant for the presenters. They didn't answer them when they should have, which amount to lies of omission in my view. I am familiar with how svelte works and thank you for taking the time to explain, as these guys had an hour to do so, and didn't bother. One of the key points that can be gleaned from your example is that there is still diffing going on, it's just in the state library instead of the template props. Additionally, svelte must then propagate changes to template proxies, which there is also record keeping for. Svelte's approach is faster for small trees and slower for large trees. It's misleading to not say so. People with much better stuff to do, like researchers and phd students, may rely on this misinformation for their work.

[@SimonLayfield]

@@jeremycoleman3282 the video is a high level overview so I wouldn't expect that they would go in to this level of detail explaining how the internals of Svelte actually work. That being said, if you are interested in getting answers to these kinds of questions then I would highly recommend jumping on the Svelte Discord server where contributors and members of the Svelte community can help.

[@kaihsiangju]

@@jeremycoleman3282 there exist a lot of metrics showing the performance between reacts and svelte online, or you can implement your own and show some metrics and data before calling others misleading. Also, if you are a real researcher, it is your job to verify the results before taking whatever is put out. Write some articles, put out charts and data and your researching methods. It is a very bad behavior to calling out others before you have something to show.