https://htmx.org logo
Powered by
# htmx-general
  • m
    🤷‍♂️ I just write all my code inline in web assembly
  • g
    if you put all the behavior in one file, than changing any behavior means that whole file needs to be downloaded and parsed again
  • m
    Right, but for a given page, if all the behavior is in that page alone, it is naturally lazily optimized, right?
  • g
    i'm not sure what "lazily optimized" means here
  • m
    What I mean is that changes to that page will not invalidate other cached resources
  • g
    sure, but that's also true for pages that reference other resources
  • g
    Oh, you mean that the filename thing makes it so that if the referenced resource changes, than the URL change is reflected in resources that reference it, so they change too.
  • g
    This is what importmaps are meant to solve
  • m
    rather so
  • m
    inlining behavior separates the concerns between functional units in the DOM, thereby sparking cold fusion
  • g
    though at first glance, generating importmaps doesn't seem much simpler than a basic bundler setup
  • m
    I call this SOCbyLOB
  • m
    Or "sock lob"
  • g
    hey react devs! htmx has code splitting built in!
  • m
    😑
  • m
    being an idiot has paid surprising dividends in life, ngl (not gonna lie)
  • g
    you've heard of css-in-js, now get ready for hyperscript-in-html
  • m
    one chess move at a time, folks
  • g
    so I looked into the etag thing, and the issue is that the browser needs to send a request to check the etag, and it doesn't send requests until the resource expires, so you can't set cache infinitely like you can with filename hashes
  • g
    although if http2 makes multiple requests cheaper, maybe we could get away with setting no-cache and having the etags checked every time
  • m
    seems like you want an out of band way for the server to communicate updated resources
  • m
    Copy code
    CACHE-INVALIDATE: <url1>, <url2>
  • m
    implies state on the server tho
  • m
    😑
  • m
    three hard problems
  • m
    OK, gonna let smarter people worry about this, we have a meme programming language to screw up
  • g
    client sends
    Copy code
    Subresource-ETag: </script.js> "3irr93rf39"
    server responds
    Copy code
    Subresource-ETag: </script.js> "mei89r9420"
    client requests new version of script.js (or maybe the server uses http push, but I have no idea what that actually is)
  • m
    Copy code
    Subresource-SHA: "4468e5deabf5e6d0740cd1a77df56f67093ec943"
    client checks against current sha, issues a
    Copy code
    SUBRESOURCES <current url>
    if not match gets list of subresource SHAs and verifies each against cache getting kinda complex 😑
  • m
    maybe a query parameter isn't such a bad idea
  • s
    ok, just for the record: I moved the event handler out of the inner fragment (see: https://discord.com/channels/725789699527933952/725789747212976259/879447087241113610) and into an hyperscript fragment: 
    _="on htmx:afterSwap log me"
    this gets called after the new content is in the dom and therefor works fine.
1...344345346...1146Latest