ah, maybe not quite

so the problem is that I don't want to replace the whole "posts" block

but load polling will replace itself, right?

nah

If you want to stick with regular polling, you can have a hidden input and use

ic-include

or does hx-swap beforeend work for it too?

yes

🤔

yeah @User right now I started doing something like this

an additional hidden element for polling

OK, I gotta get work done, I will leave you in the capable hands of Guardsman Bob

ya my basic goto is just store a variable with last loaded id and ask for id''s larger than it.

so how would I store a variable with htmx? or do I need hyperscript for that?

I mean except than a hidden element

I guess the pure htmx way is a hidden element 🙂

but you could also do polling with JS and make the htmx request view the htmx js api

I sometimes just write JS if I suspect the complexity will grow over time.

I haven't actually looked into hyperscript yet, I'm waiting for it to mature a bit first 😄

@User can you elaborate? I am trying to start a download using django (I already have a working version with django but wanted to built in HTMX for message handling). My problem right now is, that if I send something like a partial-messages.html like I usually do with HTMX, I have nowhere to put my download in the response from the django backend

could I put the download start command in the HTMX part of the response?

😑 but also 💯

Yes. This. The original page content would say something like

<div hx-get="/foo" hx-trigger="load 1s">...</div>

then the new content would include that same code somewhere in the response, like

<div hx-get="/foo" hx-trigger="load 1s">New content voila!</div>

. @mysterious-toddler-20573 wrote up a slightly longer description of this here: https://htmx.org/docs/#load_polling. However, keep in mind what @User said. Polling every 1 second is going to put a lot of load on your server. (Imagine having 10,000 people leave their browsers on that page).

So, it's a good way to get started, but if you really need real time responses (probably anything less than 30 seconds) or if you expect a substantial amount of traffic, you'll want to plan to switch to Web Sockets or SSE eventually.

I went the SSE route. It is a little more work to set up on the server, but it's not very difficult. I'm using Go, not Django, but the principles should be the same.

One more thought. There's another technique called "long polling" (https://www.google.com?q=long+polling) that is maybe a halfway step in between "load polling" and SSE. You'll find better descriptions on Google, but the basic idea is that the server simply delays sending back a response until it has something to say. It should work with the "load polling" technique described above (though I've never tried this specifically). You'll need to do some work on your server, but (probably?) not as much as setting up an SSE connection. YMMV

hm ... I can live with a download prompt if everything works well OR an error message if some failure happen

"download prompt" -- I may have misunderstood you. Are you downloading some other file, like a PDF or something? That probably wouldn't work very well in a "push" situation.

Too many end-user interactions required.

<div hx-get="/messages" hx-trigger="load delay:1s" hx-swap="outerHTML"> where "messages" is my django response with the error text in it. I am confused about the hxswap.

Ok. that makes sense.