[MainUI] Extend default standalone/item/cell widget selection

[flo-mic]

This PR extends the default widget selection for standalone, cell and item default widgets to provide better default widgets on DateTime and Number items as requested here #399 and here #1494

Basically the PR adds the following default widgets:

• oh-input for datetime on items of type DateTime (standalone, list)
• oh-input for numbers on items of type Number (standalone, list)
• oh-stepper on Number:Temperature items as well as for items with the semantic class Control or SetPoint with semantic property Temperature (standalone, list)
• oh-slider on Number:* items with the semantic class Control or SetPoint and a semantic property from ColorTemperature, Level, SoundVolume (standalone, list, cell)
• oh-toggle on Number* items with the semantic class Control or SetPoint and a semantic property from Light, Power, Energy (standalone, list, cell)
• oh-toggle on Number* items with the semantic class Switch (standalone, list, cell)

[ghys]

This looks great, I'll have a look shortly so we can eventually have it in the next release 👍

[relativeci]

Job #640: Bundle Size — 15.96MiB (~+0.01%).

44ac583(current) vs 3311fb2 main#639(baseline)

Metrics (no changes)

	Current Job #640	Baseline Job #639
Initial JS	1.73MiB	1.73MiB
Initial CSS	608.3KiB	608.3KiB
Cache Invalidation	83.71%	83.71%
Chunks	218	218
Assets	688	688
Modules	2007	2007
Duplicate Modules	108	108
Duplicate Code	1.8%	1.8%
Packages	133	133
Duplicate Packages	15	15

Total size by type (2 changes)

	Current Job #640	Baseline Job #639
CSS	856.34KiB	856.34KiB
Fonts	1.08MiB	1.08MiB
HTML	1.21KiB	1.21KiB
IMG	140.74KiB	140.74KiB
JS	9.03MiB (~+0.01%)	9.03MiB
Media	295.6KiB	295.6KiB
Other	4.59MiB (~-0.01%)	4.59MiB

---

View job #640 report View main branch activity

[ghys]

Having second thoughts:

oh-input for datetime on items of type DateTime (standalone, list)
oh-input for numbers on items of type Number (standalone, list)

I think most DateTime and Number items are really statuses, even if they're not readonly (the default), and it would be wrong to display an input UI in those cases as well as have adverse side effects (like the state description's pattern being ignored).
I'd offer the input widget by default only if the item has a Control or Setpoint semantic class - the tags are easy to add.

oh-toggle on Number* items with the semantic class Control or SetPoint and a semantic property from Light, Power, Energy (standalone, list, cell)
oh-toggle on Number* items with the semantic class Switch (standalone, list, cell)

Not sure about the rationale for these two, especially the first one. Their state wouldn't be displayed at all because the toggle doesn't allow it. If their class is Switch then it could make sense to display a toggle in all cases (not only numbers but other types accepting OnOff commands too, like Color).

[ghys]

Ok so I modified it as follows:

Cards, list items

• if the semantic class is Control or Setpoint AND the item is not read only

  • if the type is DateTime => input

    

  • if the type is Number => input

    

  • if the type is Number:Temperature OR the semantic property is Temperature => stepper

    

  • if the semantic property is ColorTemperature, Level or SoundVolume => slider

    

• if the semantic class is Switch (whatever its type but the tag should only be applied to items compatible with OnOff commands) AND the item is not read only => toggle

  

• else, no changes, fallback to a label with an appropriate action or others depending on the type

  

  

Cells

• if the semantic class is Control or Setpoint AND the item is not read only
  • if the type is Number:Temperature OR the semantic property is Temperature => knob
  • if the semantic property is ColorTemperature, Level or SoundVolume => slider
  • (nothing else because there are no input cells)
• if the semantic class is Switch (whatever its type but the tag should only be applied to items compatible with OnOff commands) AND the item is not read only => toggle
• else, no changes, fallback to a label cell with the appropriate action or others depending on the type

[ghys]

I think it's satisfactory now, and (I hope) won't cause many regressions unless the model is wrong.