The confirmation dialog box does not work correctly when using multi_calls

[il77781]

Checklist

• [ x] I updated the card to the latest version available
• [ x] I cleared the cache of my browser

Describe the bug
In the case where multi_calls is used in the action, the confirmation dialog box appears, but the actions specified in multi_calls are executed immediately when the button is clicked, and the appearance of the confirmation dialog box does not affect their execution. These actions will be performed simultaneously with the appearance of the confirmation dialog box, before any button is clicked in the confirmation dialog box.

Version of the card
Version: 4.3.0-dev.1

To Reproduce
This is the configuration I used:

type: custom:button-card
entity: switch.dmaker_p44_7a6c_alarm
name: Turn notifications on/off
tap_action:
  action: two_actions
  multi_calls: |
    [[[
      hass.callService(
       "switch",
       "toggle",
       { entity_id: "switch.dmaker_p44_7a6c_alarm" }
      );
      setTimeout(() => {
        hass.callService(
          "xiaomi_miot",
          "get_properties",
          { entity_id: "fan.dmaker_p44_7a6c_fan", update_entity: true, mapping: [{ siid: 5, piid: 1 }] }
        );
      }, 50);
    ]]]
  confirmation:
    text: |
      [[[
        if (states['switch.dmaker_p44_7a6c_alarm'].state == 'on') var vst = 'off'
        else var vst = 'on'
        return 'Do you really want to turn ' + vst + ' the notifications?'
      ]]]

Expected behavior
I would really like to have the sequence of actions specified in multi_calls executed only after clicking the OK button in the confirmation dialog box. Is there a way to achieve this?

[dcapslock]

Hi,

1. What you call multi_calls is using an indirect mechanism to run JavaScript when the action is parsed. All object templates, no matter their name, are passed at action execution time. Your mulit_calls could be foo_bar.
2. Confirmation is actually carried out by Home Assistant when it runs the action.

If you want to have confirmation before running your JavaScript you need to code this in your own way.

I highly recommend for your use case to use a Home Assistant script. I see no reason why this cannot be done with the example you have shown.

[dcapslock]

But using browser_mod.sequence and browser_mod.delay is a backend method, right?

Can be both. If fire-dom-event method used it is a Browser call method, with any browser_mod.* service called locally, and any other service sent to backend. Simple Browser call based example below. browser_mod.sequence as direct action is backend, used if sequence contains only backend services, what Browser Mod calls a Server call. You can also do advanced Browser Mod Browser calls passing special THIS for browser_id, user_id and/or entities.

tap_action:
  action: fire-dom-event
  browser_mod:
    service: browser_mod.sequence
    data:
      sequence:
        - service: <backend action>
        - service: browser_mod.navigate
             data:
               path: /dashboard

In the documentation, the indirect mechanism is not described at all, so I didn't know anything about its specific features until recently..

As confirmed in this discussion, it is a hack that may or may not work for your circumstance. This is why it is not documented, and why I describe it the way I do so others reading issues/discussions don't get confused.

This coding seems a bit difficult for me at the moment... But I will try to figure it out in time...

It's not much different than what you are doing.

If, of course, the developers consider it necessary... Still, the need to perform several actions with one click, as I understand it, is quite in demand...

As you see in this discussion, supporting multile service calls gets messy quite quickly. e.g. Where do confirmations sit? You may be interested in THIS Frontend discussion/feature request which is where the feature should be implemented.

Overall you have a bunch of options to choose from, with their pros/cons.

1. button-card hack where javascript runs at action parse time (so no Frontend confirmation etc.)
2. browser_mod where a sequence can be used (you may need to learn Browser Mod some what)
3. button-card custom javascript as shared above (you may need to understand events etc.)

[il77781]

@dcapslock, hi!

Can be both.

Yes, I understand that... As far as I understand, fire-dom-event should be used if I need any visual changes that should occur when a button is clicked, but only in a specific browser(s)... In my case, the necessary visual changes (changing the state of the clicked button) are controlled by the button-card itself (I call services that cause the entity to change, and the button-card "monitors" the change in the entity). Therefore, I specify it as follows (without using fire-dom-event):

tap_action:
  action: call-service
  service: browser_mod.sequence

That is, I am making a Server call.
Next, I specify the services:
browser_mod.sequence - as you said, this is backend as direct action
switch.toggle - the built-in Home assistant service is probably also a backend
browser_mod.delay - I'm not sure, but it also seems like a backend
xiaomi_miot.get_properties - a service provided by an integration installed via HACS. But it looks like it's "built-in" to Home Assistant, so it also looks like a backend...
So, it's as if all the executions in my above case are happening through the backend, right?
But if I used fire-dom-event and left the other lines unchanged, all executions would happen through the frontend, right?
Which option is "more profitable," for example, in terms of using computing resources, and therefore preferable?

why I describe it the way I do so others reading issues/discussions don't get confused

OK, no questions!

It's not much different than what you are doing.

I'm sure, but I still have to figure out the JavaScript syntax... )

Where do confirmations sit?

To be honest, I don't understand... Where are the confirmations actually sitting?

You may be interested in THIS Frontend discussion/feature request which is where the feature should be implemented.

Yes, I'm all for "The Rolls Royce implementation"... But as I understand it, it's about implementing tap_actions at the Home Assistant level, right? And if it's implemented at the Home Assistant level, it might also be available in the button-card, right?

Overall you have a bunch of options to choose from, with their pros/cons.

1. button-card hack where javascript runs at action parse time (so no Frontend confirmation etc.)
2. browser_mod where a sequence can be used (you may need to learn Browser Mod some what)
3. button-card custom javascript as shared above (you may need to understand events etc.)

Full functionality (with confirmations) is available in options 2 and 3. In the future (before the theoretical implementation of tap_actions) I plan to use one of them, but only if confirmations are required (I will use a hack if confirmations are not needed). Again, which of these options is the most preferable in terms of using computational resources?

[dcapslock]

Which option is "more profitable," for example, in terms of using computing resources, and therefore preferable?

All browser_mod.xxx services ultimately run on a Browser, including browser_mod.sequence. If not using fire-dom-event to make it the local Browser call, the server handles it like all other Home Assistant actions. If you then don't specify a Browser ID then all Browsers will run the sequence. fire-dom-event is mostly used as users are often callingbrowser_mod.navigate, browser_mod.notification, browser_mod.popup etc. for the local Browser. So its not about visual change, but about where the sequence actually runs.

To be honest, I don't understand... Where are the confirmations actually sitting?

I am trying to describe the connundrum of dealing with confirmations, the genises of this discussion. If a custom element were to implement multiple calls, a user may wish each to keep the Home Assistant confirmation option, and maybe a master confirmation, then maybe also return data, then decision points. Oh! I think there is a tool for that. Home Assistant script OR custom javascript. Really, I am trying to impress on you that what you think is something simple, that it is far from that.

Again, which of these options is the most preferable in terms of using computational resources?

No difference.

Also, I have thought of a fourth option which will be cleanest but take a bit for you to set up.

4. Write Home Assistant script that takes a dictionary of service calls/data. The script executes each on turn.

I would be happy to skeleton out the script if it means we can close this discussion.

[il77781]

All browser_mod.xxx services ultimately run on a Browser, including browser_mod.sequence.

As far as I understand, the browser is a frontend, right? So, running this doesn't quite align with the @RomRider belief that

only UI/visual logic should sit in the front-end and execution logic should sit in the backend.

But apparently, there's no other way in this case...

then all Browsers will run the sequence.

Do I understand correctly that this only applies to browsers that are "active" at the time the service is launched? In other words, for example, if the Home Assistant app on smartphone wasn't open at that moment, the sequence wouldn't be run in this app?

fire-dom-event is mostly used as users are often callingbrowser_mod.navigate, browser_mod.notification, browser_mod.popup etc. for the local Browser.

Nevertheless, all of the above services lead to visual changes in a specific browser(s)... I'm not arguing with you in any way, of course, everything has deeper mechanisms, I just wrote about external "signs" purely from the perspective of an ordinary user... I just can't imagine a service that doesn't result in visual changes in a specific browser(s) that would need to be run using fire-dom-event. If there are no visual changes, then what difference does it make how the service is run?

Really, I am trying to impress on you that what you think is something simple, that it is far from that.

To be honest, I don't think it's easy at all... On the contrary, I was just trying to understand why it's not easy... And I didn't insist on doing anything in particular...

4. Write Home Assistant script that takes a dictionary of service calls/data. The script executes each on turn.

Yes, this is exactly the "universal" script I mentioned above... I even provided a link to a discussion about this particular script...

I would be happy to skeleton out the script if it means we can close this discussion.

Thank you, it's not necessary, I think I can handle this on my own. I think you have many other more important things to do. I'm sorry for bothering you so much, and I didn't mean anything bad. Of course, we can end this discussion.

Thank you for your help!

[dcapslock]

So perhaps Browser Mod does not fit any paradigm correctly, but a solution. There is more than visual in Frontend, e.g. navigation. Anyhow best to leave discussion of Browse Mod.

I am pretty much completed on a template that will provide option 3 and will post it when ready in a new Show & Tell discussion.