[shobhit-seg]

Summary

• render query results in bounded 100-row pages with a 1,000-row display cap while preserving complete JSON exports
• replace nested read-only input widgets with lightweight click-to-copy selectable cells
• build query and Explorer rows in 20-row chunks and apply a per-frame UI callback budget
• enable Explorer table clipping and discard stale render chunks

Controlled benchmark

• previous 1,000x20 query grid: about 161 ms and 41,022 DearPyGui items
• new 1,500-row stored result: 0.85 ms page setup, five 1.19-1.32 ms chunks, and 2,133 displayed items

These are controlled widget-construction measurements, not production query-latency guarantees.

Validation

• 133 tests passed
• Ruff check passed
• Ruff format check passed
• compileall passed
• query and Explorer DearPyGui smoke construction passed
• git diff check passed

[abhi204]

🤖 AI pre-review — posted by Claude Code on behalf of @abhi204 (model: claude-opus-4-8). Not a formal review; @abhi204 still signs off.

No blocking issues found. Checked for bugs, security, and CLAUDE.md compliance.

Notes from the review (no action needed):

• Generation-based cancellation (render_generation, _active_table_tag) correctly discards stale row chunks on page navigation and new queries.
• Tag counters stay in sync: _render_result_page builds group tags from table_counter + 1 and _setup_results_table increments to the same value; the prior group is deleted first, so no tag collisions across pages.
• JSON export stays complete: _convert_results_to_json iterates full last_query_results uncapped/untruncated, and cell copy uses the full original_cell while only the display label is truncated.
• async_worker is wired on both TabbedQueryInterface and DataExplorer; the None fallback renders synchronously.