Skip to main content

IndexFieldNames

property IndexFieldNames: string read write

Example​

procedure ScriptEvent(var Value: variant);
begin
  DATA1.IndexFieldNames := 'CustomerCode;DocumentDate';
  DATA1.First;
  Value := DATA1.IndexFieldNames;
end;

Usage​

IndexFieldNames gets or sets an ascending case-sensitive client-dataset sort defined by ordered semicolon-separated field names.

Usage notes

Set once before deterministic traversal where possible. Use semicolons, not the commas used by CheckFields, and reacquire cursor position by key afterwards.

Errors and side effects

Missing/unsupported fields, too many key fields, pending-edit validation, provider fetch, sort/index creation, aggregate/buffer work or events can raise. A large sort can allocate and take material time. It does not edit record values or commit externally.

Additional Technical Info​

IndexFieldNames is the inherited client-dataset property explicitly exposed by Velox. It creates/selects a runtime sort from ordered field names separated by semicolons.

The example sorts ascending by CustomerCode then DocumentDate and explicitly selects the first row. It is source-reviewed and was not executed by the documentation workflow.

Declaration and syntax

Names and ordering are significant. This surface requests ascending, case-sensitive sorting; it exposes no descending/case-insensitive/grouping options. Field-based indexes do not support grouping or maintained aggregates. Setting an empty string leaves field-index mode and returns the default/no-named-index order.

IndexFieldNames and native IndexName are mutually exclusive. Setting one changes what the other reports/uses.

Active setter behavior

On an active dataset, the setter enters browse mode (which can post a pending edit), updates the cursor/provider position, fetches remaining provider data, and when the value/mode changes calls SortOnFields or switches index. It then resets ranges, checks master range and resynchronises buffers. Aggregates, if active internally, are closed/reset and buffers rebuilt.

The current logical record may be resynchronised, but its numeric RecordPosition and all subsequent traversal order can change. Do not retain sequence positions across this assignment; call First/Locate or use a business key after sorting.

On an inactive dataset, the requested value/mode is stored for later activation without an immediate sort or field validation against an open cursor.

Performance and concurrency

Building/switching the index depends on record count and key width; subsequent ordered access can be faster. The index/cursor is shared mutable state and unsynchronised.

External references

Created 2026-07-15