https://htmx.org logo
Join Discord
Powered by
# htmx-general
  • l
    Right, but that's not driven from user actions, right?
  • f
    True. At least, not directly.
  • l
    Anyway, I don't see the emergence of Turbo as a negative thing, I see it as a positive development. It makes it easier to sell the architectural choice. Then, it comes down to what's a better fit for the given project.
  • l
    Ohhh. That's nice, they have a iOS & Android shell applications that simply delegate the work to the backend. 😉
  • m
    yes, please do
  • m
    Thank you, I will think about adding some wording about that.
  • m
    Turbo looks about like what I expected: extremely polished and well presented, but much more framework-y and magic than htmx
  • m
    I mean, HTML, really? In 2020? Like H T M L? Seems like an antipattern. Also, separation of concerns.
  • m
    8k followers mfw
  • m
    😑
  • m
    It's all good. The turbo stuff is good news since more people will listen to DHH than to me. HTML on the wire is making a comeback, and hopefully htmx is part of that and is useful.
  • g
    "Yes, true pure technical web dev probably does require HTML."
  • m
    If you can explain to me how you can be REST without JSON, I'll listen but I have to say you are only going to make yourself look silly
  • l
    @User So, I think our backend is going to be almost completely WebSockets. How hard would it be to tunnel all HTMX requests / responses over WebSockets?
  • l
    There are downsides to going with websockets... basically, your client is literally tied to a particular backend, there's no good way to re-use any HTTP infrastructure with regard to caching, routing, traffic analysis, etc.
  • l
    On the other hand, WebSockets have become the primary way that interactive data science applications are built. This gets down to the frame level; in HTTP you have to send the Content-Length unless you're going to be Chunked. While, it's a flag on every web-socket frame if the next frame continues the same content.
  • l
    Anyway, I looked a Turbo, I don't see why I'd want to use it over HTMX. It seems like many of the interaction patterns (settle, etc.) aren't available. Also, I see DHH growing a very tight ecosystem, and I like that others are using HTMX by itself without having to buy-in to the entire suite. Being able to use Alpine, in particular, seems like it'll be a requirement we'll have.... there are so many external web components we'll need to use.
  • l
    I love how people use "separation of concerns" to push their favorite clustering of features. I've run into cases where there's a good many ways to factor a problem, and well, sometimes a simple monolith works great... but, for "separation of concerns" to pull out a micro-service or something that is actually way more coupled to everything and hardly separate. When I hear about "separation of concerns" I think about "you deliver your org-chart" since it often means, "I want a small boundary around what I am responsible for delivering, because those others always mess it up and I don't want to be called at midnight".
  • g
    https://twitter.com/technosailor/status/1341478444136361985?s=20 really just said "this is bad because it's easy" huh
  • l
    Oh horrors... developers staying "in their comfort zone"
  • g
    you might think web dev is about making websites, but it's actually about flexing our galaxy brains
  • m
    We have WebSocket support, but I can't say it is battle tested, so there might be some bumps. I can say that I'll help get it working for you.
  • l
    @mysterious-toddler-20573 Cool. I'm now finished with https://github.com/clarkevans/HypertextLiteral.jl and am now moving onto the TodoMVC, but I think I'll have to use Websockets out of the gate.
  • l
    I see this workflow.... SQL -> HTML.
  • l
    Oh dear.... you mean I forgot the middle tiers and other abstractions?
  • m
  • m
    should we add a 
    hx-preload
    option for htmx?
  • m
    where we issue a get on hover-ish and cache the results
  • l
    Ah, make a request, get the data there, by guessing where users will go.
  • m
    It's all hover-based, makes sense. I looked at the instaclick.js codebase and it's a bit crazy, but I think I get the general idea and it wouldn't be a ton of code in htmx
1...106107108...1146Latest