What's Changed
Compute-pool routing and usage attribution for the DeltaForge ADBC driver.
- The Arrow-native driver now participates in DeltaForge's compute routing. Every query carries an
x-deltaforge-nodeselector: when you have pinned a compute node the request goes straight to it (direct-dial), and when no node is selected the server picks one for you (load-balancing). The driver behaves the same whether you point it at a single node or a routed compute pool. - Requests are tagged as ADBC traffic (
X-DeltaForge-Client: adbc), so DeltaForge's per-table and fleet usage dashboards record when your data lake was reached through the ADBC driver. - Versioned to the 1.0.8 platform release.
These are two new outbound request headers; Arrow result handling, type mapping, and schema are unchanged from 1.0.610.