CSharpDB v2.8.1

v2.8.1

Admin Query and View Responsiveness

• Removed the blocking upfront COUNT(*) requirement for ad-hoc SQL and view browsing in CSharpDB.Admin, so first-page results can render immediately.
• Added unknown-total paging state for query and view grids, including improved row-range status when the full result count is not yet known.
• Added a forward-only direct query cursor for view paging so sequential next-page navigation can continue without re-running the full view query.
• Improved filtered view behavior by pushing simple outer predicates into the underlying view source instead of evaluating them only after the expanded join/view result.

SQL Planner and Execution Improvements

• Added simple-view outer predicate rewrite and broader inner-join leaf predicate pushdown so selective filters can reach the correct base-table source earlier.
• Extended these pushdown rules across qualifying inner-join chains instead of limiting the optimization to a fixed join count.
• Normalized identity arithmetic join predicates such as x + 0 = y so they use normal join planning instead of falling back to slow nested-loop behavior.
• Extended compact scan fast paths to cover simple projected LIMIT/OFFSET queries, including SELECT * single-table scans.
• Extended fast indexed lookup paths so LIMIT/OFFSET is preserved on star, covered-projection, compact-payload, and generic projection indexed plans.
• Tightened SQL parsing so JOIN syntax without the required ON clause is rejected as invalid instead of reaching execution.

Benchmark and Validation Updates

• Reduced benchmark baseline storage to a single focused validation snapshot and updated guardrail configuration to use that single baseline consistently.
• Renamed stale benchmark labels and baseline rows so guardrail comparisons match the current benchmark surface after the join and scan fast-path changes.
• Refreshed the benchmark validation pass and confirmed the current release guardrail comparison is clean: PASS=184, WARN=0, SKIP=0, FAIL=0.