FireDAC Client

When the server exposes FireDAC datasets, the client can fetch them into live TFDMemTables, let the user edit them, and post the changes back as a delta. This gives you a near-classic data-aware experience over REST. The component is TMARSFDResource (MARS.Client.FireDAC.pas), installed by the MARSClient.FireDACDesign package.

Setup

Drop a TMARSFDResource, link it to an application, and tell it which local datasets correspond to the server's:

FDResource.Application := App1;
FDResource.Resource := 'orders';              // a TMARSFDDatasetResource on the server
// ResourceDataSets maps server dataset names to local TFDMemTables

TMARSFDResource holds a collection (ResourceDataSets) pairing each server-side dataset name with a local TFDMemTable. Set these up at design time (the component editor lists the names) or in code.

Fetching data

A GET populates the linked mem-tables:

FDResource.GET(
  nil,
  procedure (AStream: TStream)
  begin
    // OrdersMemTable and ItemsMemTable are now filled and active
    Grid1.DataSource.DataSet := OrdersMemTable;
  end,
  nil);

Because the server can return the compact FireDAC wire format, transfers are efficient and field metadata (types, constraints) is preserved.

Sending changes back

Edit the mem-tables as usual (FireDAC tracks the changes). A POST sends only the delta to the server, which applies it with ApplyUpdates and returns a per-dataset result:

// user edited OrdersMemTable / ItemsMemTable ...
FDResource.POST(
  nil,
  procedure (AStream: TStream)
  begin
    if FDResource.ApplyUpdatesResults.AllOK then
      ShowMessage('Saved')
    else
      ShowMessage('Some rows were rejected — see error details');
  end,
  nil);

The server returns an array of TMARSFDApplyUpdatesRes (applied count and any per-row errors per dataset), which the client exposes so you can report or reconcile failures.

Single-dataset resource

For the common one-dataset case, TMARSFDDataSetResource binds a single TFDMemTable and exposes convenience properties like Filter and Sort (sent as query parameters) and flags controlling whether to send deltas. The usage pattern is the same: GET to load, POST to save.

End-to-end shape

[Client]  TMARSFDResource.GET  ──►  GET /rest/default/orders
                                     server: TMARSFDDatasetResource.Retrieve
          local TFDMemTables  ◄──    JSON / FireDAC binary (datasets)

  user edits rows (change tracking) ...

[Client]  TMARSFDResource.POST ──►  POST /rest/default/orders  (delta)
                                     server: ApplyUpdates(datasets, deltas)
          ApplyUpdatesResults ◄──    [{ dataset, result, errorCount, errors }]

See the server side in FireDAC & Datasets and the working ConnectionPoolingProject demo.