https://htmx.org logo
Powered by
# htmx-general
  • m
    need to think more about it
  • g
    i want the url to represent the state of the application, not the last query made
  • g
    i think that's the most important functionality of the url, to be able to represent the current state the user is trying to share
  • m
    Right, but from a web perspective the request made is the state. HATEOAS
  • m
    You can use the 
    HX-Push
    response header to clean out those parameters...
  • m
    On the server side, I just realized
  • g
    it's irrelevant to me server side if those parameters are there or not; i'm thinking purely ux, when i click a page and my url gets weird and when i try to share it it doesn't look like something that represents what i was doing
  • m
    Right, I get it.
  • g
    i treat empty params the same as non-existent params (as servers should =P)
  • m
    Is the 
    HX-Push
    header a viable solution?
  • g
    sorry, i don't mean to keep repeating myself, sometimes i'm just not sure if we're talking about the same thing because i don't have all the terminology you have
  • m
    No worries, I get it
  • m
    the pipe is narrow over a small text box
  • g
    uh, i'd have to try it specifically, not sure what the result or implementation of that is yet
  • m
    You'd construct the URL you want on the server side and return it via that header. Could be pretty general (iterate over parameters and filter out ones that are empty)
  • m
    I'll think more about a client side solution as well
  • g
    that makes sense. i'll try it out later when i get some time.
  • m
    👍
  • g
    i'm doing my demo next thursday, i'm excited.
  • g
    building reusable components on the server has been an interesting experience. it's been many many years since that was a common thing to do in django, so support is not as complete as it could be
  • g
    https://gitlab.com/mekhami/htmx-demo/-/blob/main/stocker/ledgers/templates/ledgers/move_modal.html i'm doing some shit here with hyperscript that i'd like to remove -if its possible-. i was trying some stuff with html5 validation and i couldn't get it to work the way i expected. basically for my demo i want to minimize the use of hyperscript, since my premise of the demo is 'write better apps, and fuck javascript'
  • g
    i guess one thing i could do is just refetch this whole modal on 
    keyup changed delay
    and let the server handle the validation state
  • g
    but with a multi-input form, if i enter an invalid value in form 1, and refetch the whole form, it's also gonna show 'required' errors on the other form fields (because i had to post the whole thing)
  • g
    so i think i really do want individual field validation
  • s
    ive actually been thinking about forms and htmx for an app im building and I made the following conclusions:
  • s
    1) making forms fancy and complex and full of state changes is likely a waste of time and will jsut confuse users, I have settled on a workflow that is: open form -> submit -> (fix errors -> submit)*
  • s
    2) I do as much client side validation as possible, the html standard has good support for validation these days and using the browser to validate gives the best user feedback.
  • s
    3) server side validation is only triggered in cases when people try to cheat the system or i made a programming error. So I simply return a 403 with a hx header to trigger some simple overlay informing the person about the problem.
  • s
    I haven't really followed the talk so no idea if any of this applies to you,m but I pondered forms for a long time an finally resolved to using in browser validation and if the server gets invalid input then I just display a simple message and possibly log it as a bug to investigate.
  • s
    In other projects i see devs spend soo much time on forms and validation, when most of the code is handling states that are very unlikely to be encountered.
1...176177178...1146Latest