I can’t get this to work. I sort of got it once, but the value in a target reference column had the right text but was not linked back to the source table.
Now, nothing works for the action button.
Here is the error:
"ActionButton" cells should contain an object with keys "button", "description", "actions". Missing keys: "button", "description", "actions"
Here is the formula from the action button:
{"description" : "Schedule a task", "button" : "Schedule", "actions" : [["AddRecord", "Details", None, {"Item": Routine_Maintenance.lookupOne(id=$id).cat}]]}
Looks like the dictionary has all 3 keys: description, button, actions. (Yes, there is an equal sign before the dictionary literal, but that won’t copy/paste.)
I’ve edited the formula multiple times to hopefully get it right. Clearly, I didn’t. I delete the widget to add it all over again because you only seem to parse the formula in the ActionButton column once when it is added. I got it work (incorrectly) once. Now it just never works at all.
I know you appreciate it when users show their undying gratitude for the great work you do, but sometimes the frustration is a bit much.
- Python formulas are essentially arbitrary code even when wrapped in a dictionary literal. There has to be a way to introduce cosmetic line breaks and to make the edit box larger. Even Excel allows both.
- Need a lot better documentation of the objects and functions that Grist exposes to Python.
- Widgets are fundamentally designed to be big rectangular areas that can hold a table, a card, a series of cards or a plot frame. How does a single little floating button get a huge rectangle for which the context is unclear? The button might apply to the document as a whole, but it might also apply to a row, in which case it should be the rendered value inside a column – not unlike a choice item.
Sorry, but 3 hours later for something like this has me wondering if you should release features in such raw state just to get marketing credit for having the feature. Don’t you want the feature to be usable before you release it?
So many things in the design of this product are very well thought out from the beginning (like granular permissions). As you scramble to keep up in the feature battle don’t succumb to “featuritis” just adding things by cramming them in. Better to delay and get it right than get it wrong–that’s worse than no feature at all. Hard discipline when competition in this category is truly frantic.