https://htmx.org logo
Join Discord
Powered by
# htmx-general
  • h
    Sure.. using something like this:
    Copy code
    <form method=“POST”>
<input type=“text” required pattern=“[0-9]*” hx-post=“/validate-something” hx-trigger=“click”>
</form>
    Html5 validation doesn’t kick in and the required and pattern attributes are ignored, allowing the post request to go through. In your thread on issues you debated this with someone else and said that you would expect this behaviour rather so that server side validation could happen without browser interference, which I agree is the right approach. But I can’t get the html5 validation to kick in at all. Any ideas?
  • h
    Sorry that should have been keyup changed not click
  • h
    The correct hx-trigger I’m using is ‘keyup changed delay:1000ms, blur delay:100ms’
  • h
    In either case it wont check for the required attrib and the empty value is submitted
  • h
    I’m guessing this is because only the submit event calls checkvalidity?
  • g
    I ran into the same thing and yes, it needs to be the submit event in order for the browser to check validity before posting.
  • h
    Were you able to find a way around it?
  • h
    I know some hs or alpine can do it fairly easily but I’m asking here because of the GitHub issue and the [valid] filter which would make life far less … complex
  • g
    I had to move on with just waiting for the submit, but I'm going to come back using a similar approach to this post http://hernantz.github.io/inline-form-validation-with-django-and-htmx.html
  • e
    Is there a way to have htmx set hx-target to one value when hx-trigger is input, but a different hx-target when the hx-trigger is submit? Trying to work out the best way to get around some user experience issues I'm hitting. Another potential solution I tried is having in the form with the hx-target I want with hx-trigger being click, but when I do that the form POST data doesn't include the button's value.
  • e
    Easiest is for me to just replace the whole form, but if I use hx-trigger of "input delay:1s", and a user is in an input with cursor at the end of that input's content, then when the content div is replaced, the cursor is moved to the start of that input.
  • m
    You can put the hx-get or whatever on two different elements and have them target different things
  • m
    You could also use the morphdom extension which should do a better job of preserving focus and caret position
  • h
    so what's the deal with calling things "microservices"? I get the gist that they're supposed to be loosely-coupled but isn't that just generally good programming practices?
  • m
    a microservice is deployed separately
  • m
    often runs "serverless", which, in my experience, means: "I don't know how this runs"
  • m
    but, in fact, if i understand correctly, means: "runs in a vm-based server managed by amazon in an opaque manner"
  • h
    whereas a monolith is deployed as a single "instance"?
  • m
    correct
  • m
    imo microservices conflate (at least) three things: proper system decomposition, deployment and scaling. A lot of people appeal to Conway's Law regarding them: each department or developer gets their own independent development environment, through deploy.
  • m
    suffice to say, I am not a fan
  • h
    thanks for the explanation. seems like the notion of microservices is pretty vague
  • m
    yes
  • m
    i don't mind macroservices
  • m
    a database is a macroservice
  • m
    something like a system that is responsible for all email delivery, fine
  • m
    but microservices seem to be addressing an admitted problem: deploying multiple changes across an org is hard, by making it really hard to refactor across the code base.
  • m
    and the deployment, monitoring, etc. all ends up getting pretty crazy anyway
  • m
    anyway, I'm not a fan, but I am old and grumpy
  • m
    not to say it's always inappropriate, but I lean monolith by default
1...788789790...1146Latest