Hi.
I am attempting to make a reusable grid using a passed ri! of type Any with a cdt type being passed into it.
Optionally I am thinking about passing in a list of column names so I can specify the order the grid displays columns in.
From there I wanted to do a foreach generating the N grid columns using field names and types grabbed from the passed cdt..
1.I don't know if there is a way to get a list of all the fields a cdt type has.
2. I am unsure if there is a way to get their types.
Any tips?
Discussion posts and replies are publicly visible
roberts0186 said:I don't know if there is a way to get a list of all the fields a cdt type has.
a!keys() -->
FWIW, this sounds like a neat exercise but I don't expect that this level of "reusable grid" will be worth the effort you put into it, in a production-usable sense. The read-only grid component is already easy enough to configure that it can basically just be designed as-needed, or if you have a use case where you need a bunch of different interfaces to use a similar one, then you can set up a reusable one under that slightly-narrower scope.
There are other bits of functionality I want to add in that increase its worth if reusably programmed.
Did not include them here was I wanted to keep the question simple.
I support Mike point of view. Creating a reusable component this close to the original will probably not help much.
Over the years I developed quite a few reusable components. In case you are interested, I would be willing to support you. Which functionalities do you plan?
I was brainstorming filter designs. Most of these are oriented around uncluttering filter usage without forcing the user to drill into hidden filters. When i say a filter I imagine something in the format below.
[CreatedBy]:[____enterTextHere_________________]
1. A filter that toggles the filtered field based on a set of autogenerated buttons in a line above your column names, named after the column names. See below
[CreatedBy] [CreatedDate] [Serial]
[______Enter text here_____________]
-I wasn't a fan of how cluttered things get with a dozen _Enter text heres__ normal filters in mass.
-Even with hiding them when not in use, I feel they either take too long to drill into or too many are shown at once with larger queries.
-Buttons would reduce this space usage and allow faster filter choosing and cleaner space. This could cover about twice the number of column names in the space a normal filter uses.
- Design decision with this is only one thing is filtered on at a time.
2. Most users probably only filter on a few columns. A way to let users choose their own configuration of normal filters might be useful. Add on filters by giving them a multi selectable dropdown with column names to enable/disable.
-I would like the dropdown autogenerated.
Some features seem too complicated as I type them out like letting a user create a preset that exists each time they log in. Plus I feel they would require creating tables for the info and would definitely stop the generic-ness of the rule I has wanted.
Ex.
(Kyle sets it up so he sees ID and FinishDate filters) (Marie sets her preset to see filters for ID and CreatedBy)
Anyway I think this is all a bit beside the point. Came here looking for ways to extract column info from things like datasubsets to automate the above. Thats what I want tips on here.
Do you happen to know how to use this with a datasubset?
I had thought i could just use the .data field, but apparently despite grabbing an individual dictionary(which documentation says works) from it, key function doesn't seem to work on it.
roberts0186 said:map function doesn't seem to work
What do you mean by "map function"? Do you mean a!keys()? Or are you trying to do something else? Can you post a code snippet, and share what you were hoping for (versus what you're achieving)?
Sorry. yes the a!keys function on a datasubset.
Example is a!keys(value:ri!mydatasubset.data[1])
value being of the format
[isActive:true,modifiedBy:fin.financeadmin,createdOn:6/21/2021 3:19 PM GMT-05:00,createdBy:finley.financeadmin,purchaseOrderCategory:Haste,clientPoId:To,purchaseOrderId:105]
Returns error for the above line as "The passed parameter(s) are of the wrong type. Received the type Any Type"
I think my original suggestion was that you could add a type (or just type id) as a parameter of your reusable component, and call a!keys() on that type reference inside the component.
For example if you pass a relevant type number (integer), into this code, you get the requested keys: a!keys(cast(6590, {}))
(6590 is an arbitrary CDT type number in my system, sub out for one of yours, which you can discover with the "typeof" function)
The keys function should work on a dictionary. Is it possible that the data you provided is empty? The biggest issue I see with this paradigm is that it is entirely dependent on the data that is provided, so if for any reason there is no data, it won't be possible to generate any columns either.