https://htmx.org logo
Join Discord
Powered by
# htmx-general
  • b
    > some special HTML element with the ID "htmx-extras" as the last element in the . This sounds like it would fit nicely as a special case of OOB-swap
  • m
    I got a bug and implemented a playground verison of the algorithm I have in mind: https://github.com/bigskysoftware/htmx/blob/3506ecfb6a12c995871c2c02b35a481825930a10/test/merge-head-scratch.html#L22
  • m
    basically a modification of the existing logic to get the title tag out, coupled w/ the child-syncing logic @brainy-ice-92385 suggested
  • m
    I use 
    outerHTML
    , which is grug-brain but works
  • m
    I played around with it locally and it appears to work pretty well. This is an important enough change that I'm going to hold off on the next release of htmx to see if I can convinced myself this works, and maybe include it.
  • m
    We will need events for all the add/removes that occur, but I like the look of this. Coupled w/ idiomorph, should be a nice combo.
  • b
    Just had an interesting thought which is to send a remove event to a script tag itself. Then de-init code could listen for removal event on its container script tag. Have't thought it through much but it's fun
  • m
    Yep exactly we are on the same page
  • m
    now I'm wondering how to support this
  • m
    Right now we support a title tag in the content
  • m
    it seems to me like people would want two different use cases for a 
    <head>
    tag: * You are getting a full html document response (e.g. a boosted link, or you are using hx-select) and you want to synchronize the current head element w/ the new one * You are returning a partial, but you want to jam some additional stuff up in the 
    head
    element
  • m
    thinking if you mark the head as 
    hx-swap-oob='beforeend'
    we will do the latter behavior
  • m
    does this sound right?
  • m
    also want a way to mark stuff as preserved, in case it's added dynamically
  • b
    Struggling a bit with the swap oob case. If they're merged, does that mean if you want to make any change to the head, that you will have to send an entire head that is mostly the same to prevent things being wrongly removed? Does that make sense - I can rephrase
  • b
    For that case, it would be useful if you could nest some chunk of the head in a semantically uninteresting element like a div, just to mark that as the "volatile set of stuff that we might want to merge changes into" and keep stuff that never changes, like the meta viewport or base css out of that. But that's probably too complicated an idea (is it even possible, idk)
  • b
    On second thought your marking as preserved idea makes a lot more sense heh
  • m
    I'm thinking the "here is a whole page, replace the head" case vs. the "here is a partial bit of HTML, and some stuff to jam in the head" case
  • b
    Yes those two cases make sense. For the latter, you could just make a special case of "If you provide a tag at the top level of a response, we will treat it as this special case of oob-swap with the special merge"
  • m
    so "If there is an in the response, we sync, if not, we merge?"
  • m
    maybe a little too magic?
  • m
    Just wondering if we want to give people something explicit
  • b
    I think a at the top level would be a non-magic signal that this is a fragment, not a whole page
  • b
    at the top level having meaning does not seem too magic to me
  • b
    like, obv it is magic but it's not too magic IMO because those things align with the spirit of HTML. Maybe a top-level htmx config could opt in or out of that behavior
  • b
    Put another way, what else could those elements as top-level mean?
  • c
    merge seems too much complexity...
  • c
    not to implement but just the concept of htmx i general... in my mind it is supposed to be stupid simple
  • c
    I know that may clash with people seeking to write pure html instead of generating it but idk I think htmx may get to complex if it plans to support merging (which is really avoiding templates or other generation mechanism)
  • c
    i dont think the prescense or absence of html tag should do anything at all
1...848849850...1146Latest