Tables and fields
A table's columns are a closed set of six types, each cast on the way in, plus the indexes that make one row findable.
A table is defined once, in Xuda Studio, and every place that shows its data, the Studio grid, the dashboard's Data tab, a Get Data program, or a direct API call, reads the same field definitions.
This page covers fields the way you meet them building and browsing a table. For the complete property list, indexes, and how a Get Data or Set Data program points at a table, see Table in Building Blocks.
Six field types, nothing else
Every field on a table is exactly one of six types. Studio shows each with its own icon in the
grid header rather than a text badge, since a column is narrow and a field id like ORD_Total is
already most of the width:
There is no seventh type, and no way to add a custom one. A field that does not fit these six,
a rich-text block, a file, a relationship to another table, is modeled as a string (an id,
a path, a JSON-encoded blob) or an array/object of them; there is no separate first-class
column type for any of those on a table.
What each type actually does with a value
Both dbs_create and dbs_update cast the value you send through the field's declared type
before it is stored, so it is worth knowing exactly what each cast does rather than assuming a
familiar meaning:
string
Anything that is not already a string is converted with a plain string conversion. A blank or
missing value stores as "".
number
Number(value) is stored. A value that does not parse to a finite number falls back to 0
rather than rejecting the write. A blank or missing value also stores as 0.
boolean
Only a fixed list of tokens reads as true (Y, 1, true, TRUE, ON, YES), and another
fixed list as false (N, 0, false, FALSE, OFF, NO); anything else, including a blank
value, is stored as false.
object
A string is parsed as JSON5; a value already shaped as an object is stored as-is. A blank or
missing value, or a string that does not parse, stores as {}.
array
A string is parsed as JSON5 first, and if that fails, evaluated as a JavaScript literal instead;
a value already shaped as an array is stored as-is. A blank or missing value, or a string neither
parse handles, stores as [].
date
There is no dedicated date handling on the way in: whatever string you send is stored exactly as
sent. Pick one format, ISO 8601 (2026-09-25T00:00:00.000Z) is the safest default, since nothing
in the platform normalizes a date field for you, and a table with mixed formats sorts and filters
inconsistently.
Sending a value that cannot become a valid number for a number field does not fail the call, it
silently stores 0. Validate a numeric value yourself before writing it if a 0 would be wrong.
Naming fields
There is no enforced naming rule beyond field ids being unique on the table, but every table
already in a Xuda project follows one convention: prefix every field with a short code for the
table itself, so ORD - Orders produces ORD_Id, ORD_Total, ORD_Status. It keeps field names
unambiguous once a query, an expression, or an import file has to reference several tables at
once.
Marking a field mandatory
A field carries a mandatory flag you set when you define it, alongside its type and comment, to record that a row is not meant to be saved without a value there.
Tables have their own active/archived state
Separately from any single row being archived (soft-deleted, see Querying data over the API), a table itself can be active or archived as a Studio node. The dashboard's Data tab lists a project's tables as cards, switchable between Active and Archived, each card showing the table's name and its current row count. Archiving a table here is a Studio-side action on the node, it is a different operation from archiving rows inside it.