วีดีโอ

We try to provide a better experience and much more possibilities over time than JS libraries

We will try to refatctor the ui and add more components to be better in terms of ui too. thank you for your support

@ thank you for authoring a very useful library that could potentially be widely adopted

@@norbertocammayo2334 ♥♥

I think it wouldn't, however the git diff should be enough to sort this out by yourself

@@QuentinDurot Got it. So the onus is on the library user to merge in upstream changes to generated components.

Hello, dear friend. I put a lot of thought into this during development. The generator shows you the diff, and as a developer, you need to manage itوthere's no other way around it. Even if there wasn't a generator and you used an independent plugin, you would always have the latest version.

However, we followed a unified approach across the entire component and will continue to improve it in the future. Each component has a property named class, which essentially serves as your custom class, so there’s no need to modify the component itself.

Thanks for the feedback and the suggestion! I'll consider it in the future. In the mean time, I have covered live_svelte which has a very interesting and productive developer experience.

Thanks for the very kind feedback, and welcome to Elixir and Phoenix!

Thanks, and welcome back to Elixir!

Htmx and LiveView JS commands are similar, but different. While they both use HTML attributes to provide frontend and backend interactions, Htmx attributes are directly written and perform API calls. On the other hand, LiveView JS commands are written in Elixir and serialized to an HTML attribute by the LiveView renderer. These commands don't perform API calls, instead relying on the websocket connection to the corresponding server process and receiving compact diffs in response.

@CodeAndStuff Thank you very much for explaining this. Appreciate your efforts. Then this is closer to SignalR

Same here. Subscribed !

I’m glad I’m not the only one! Font designers put so much work into variants all for us to just use the default and move on.

Hooks are able to push to the server through the pushEvent method: hexdocs.pm/phoenix_live_view/js-interop.html#client-server-communication

Thanks! I found out about the modal removal from core_components while researching this video. It turns out modals won't be used in UIs generated by the phx mix tasks, so this PR removed the modal when changing the generator: github.com/phoenixframework/phoenix/pull/5900

I'd highly recommend reading through the IETF draft for OAuth 2 Security Current Best Practices - it explains those more and what threat vectors exist. In decentralised landscapes like say Mastodon* and the Fediverse, there's also an initial flow known as Dynamic Client Registration, and a new internet draft I'm co-authoring called Client ID Metadata Documents, which tackle obtaining a client_id and potential client_secret without having to register with the authorization server ahead of time. *Mastodon doesn't currently implement DCR, but something similar to it.

Looks like it's part of an effort to simplify the code generated by mix phx.gen.live. github.com/phoenixframework/phoenix/pull/5900

It just simplifies the visuals for teaching.

That's a bit of a tricky one. There's some documentation on optimistic UIs in the liveview docs (here: hexdocs.pm/phoenix_live_view/syncing-changes.html), but it doesn't directly cover the optimistic create use case. If you strictly want the exact same behavior as that tutorial-updating the UI with a page-local state-you probably should use something like Live Svelte or build a traditional SPA (but you can still use Phoenix Channels with it to make the comms go over a websocket and enable server-to-client push). Technically, this could be written as a phx hook, but I imagine it'd be more complicated than it would be worth. If your requirements are a bit more flexible, look at async assigns (docs: hexdocs.pm/phoenix_live_view/Phoenix.LiveView.html#module-async-operations). In my experience, most of the time spent waiting for a create operation isn't a result of a slow network link. With async assigns, you can create a placeholder assignment on the socket with a loading and error state. The server can immediately push a patch to the page to show a loading indicator or optimistic item in a table, then when the processing finishes it can patch to look like a complete entry. I'm considering making a video on async assigns in the future because it's a pretty interesting solution to things like this. Let me know if you end up pursuing any of these paths!

@@CodeAndStuff Thanks so much for the reply and info. I will look into the info you have provided. I would love to see a video on the async assigns. I am wondering if there are ways to make live view sites feel as snappy as SPAs. I find navigating live view sites a bit slow compared to spa.

Thanks! I'll consider it in the future

Thanks for the idea!

I'm glad you've enjoyed the videos! It depends. I'd choose livesvelte or inertia if the project strictly requires a library that only plays well with a framework like Svelte or React, or if the team working on it has deep experience with those frameworks and timelines are tight. The best stack is the one that a team is most productive with, and tradeoffs are unavoidable. That being said, you can get a ton out of the box with LiveView JS-bailing out to hooks when / if needed. I'd definitely recommend trying out a LiveView-first approach and seeing if it meets your needs, knowing you have options.

x.com/rootcert/status/1836035549377003551

github.com/phoenixframework/phoenix/pull/5900

You caught me!

Thanks! I’d love to write Elixir at work too, but companies love their JS, Python, Ruby, etc… Some day Elixir will catch on!

Thank you! I've been having a lot of fun sharing projects with this community. Comments like these keep me motivated!