aminghadersohi commented on PR #44575:
URL: https://github.com/apache/superset/pull/44575#issuecomment-5797453111
Measured all **72 registered tools**, not only chart tools, comparing the
pre-change serializer at `f8f293d266` with `90fe26612e`. Same registered tool
models/config on both sides; complete serialized entries including metadata;
compact JSON (`ensure_ascii=False`, `separators=(',', ':')`), UTF-8 bytes, and
**tiktoken 0.14.0 / cl100k_base**. Tokens are BPE estimates, not
Claude-tokenizer counts.
**Totals (sum of individual entries): 531,433 → 348,099 bytes (−34.50%);
126,042 → 81,354 tokens (−35.45%).** Four tools shrink, two are unchanged, and
**66 grow**. Every increase is identified by its positive deltas below. The
additional large win outside the original three is `update_chart_preview`:
−49,943 bytes / −12,275 tokens.
The increases total +12,210 bytes / +3,421 tokens across those 66 tools. The
largest individual increase is `manage_native_filters`, +706 bytes / +210
tokens, ending at 6,544 bytes / 1,438 tokens. These are small absolute
increases from retaining nullable alternatives, `$defs`/`$ref` wrappers, and
reference-sibling metadata that the old pass discarded. They are accepted costs
of preserving the original published schema, not repeated-model expansion;
reinstating the old simplification would lose fidelity. No growing entry
exceeds 6,544 bytes or 1,438 tokens, so no material delivery-size regression
was found and no additional runtime change was needed.
Added `test_tool_inventory.py`: 72 independent byte/token budget cases, 72
complete-schema/non-mutation checks, and an inventory-completeness check so new
tools cannot silently escape coverage. **207 inventory/search tests and 178
execution tests pass**; all applicable branch-file pre-commit hooks pass.
Injecting schema growth into `get_database_info` makes its budget case fail as
expected.
<details>
<summary>Full inventory: every tool, including all 66 increases</summary>
| Tool | Bytes before → after | Δ bytes | Tokens before → after | Δ tokens |
|---|---:|---:|---:|---:|
| `add_chart_to_existing_dashboard` | 1,259 → 1,372 | +113 | 281 → 312 | +31
|
| `apply_dashboard_filters` | 2,553 → 2,789 | +236 | 560 → 625 | +65 |
| `create_dataset` | 1,491 → 1,648 | +157 | 321 → 365 | +44 |
| `create_theme` | 920 → 989 | +69 | 203 → 221 | +18 |
| `create_virtual_dataset` | 2,887 → 3,536 | +649 | 618 → 792 | +174 |
| `delete_chart` | 877 → 946 | +69 | 192 → 210 | +18 |
| `delete_dashboard` | 878 → 955 | +77 | 188 → 206 | +18 |
| `duplicate_dashboard` | 1,645 → 1,728 | +83 | 340 → 359 | +19 |
| `execute_sql` | 1,791 → 1,970 | +179 | 377 → 429 | +52 |
| `find_users` | 1,319 → 1,384 | +65 | 301 → 320 | +19 |
| `generate_bug_report` | 2,040 → 2,426 | +386 | 449 → 559 | +110 |
| `generate_chart` | 97,013 → 49,531 | -47,482 | 23,543 → 11,849 | -11,694 |
| `generate_dashboard` | 3,047 → 3,296 | +249 | 637 → 709 | +72 |
| `generate_explore_link` | 96,581 → 49,230 | -47,351 | 23,478 → 11,806 |
-11,672 |
| `get_annotation_layer_info` | 712 → 803 | +91 | 158 → 181 | +23 |
| `get_chart_data` | 2,526 → 2,709 | +183 | 554 → 608 | +54 |
| `get_chart_info` | 3,279 → 3,462 | +183 | 721 → 777 | +56 |
| `get_chart_preview` | 3,027 → 3,300 | +273 | 661 → 744 | +83 |
| `get_chart_sql` | 1,780 → 1,905 | +125 | 390 → 428 | +38 |
| `get_chart_type_schema` | 701 → 701 | +0 | 149 → 149 | +0 |
| `get_compatible_dimensions` | 1,236 → 1,385 | +149 | 242 → 281 | +39 |
| `get_compatible_metrics` | 1,227 → 1,370 | +143 | 243 → 282 | +39 |
| `get_dashboard_data` | 2,188 → 2,295 | +107 | 484 → 514 | +30 |
| `get_dashboard_datasets` | 911 → 998 | +87 | 195 → 218 | +23 |
| `get_dashboard_info` | 2,818 → 2,953 | +135 | 609 → 648 | +39 |
| `get_dashboard_layout` | 1,356 → 1,467 | +111 | 284 → 314 | +30 |
| `get_database_info` | 1,222 → 1,299 | +77 | 277 → 298 | +21 |
| `get_dataset_info` | 2,126 → 2,201 | +75 | 507 → 528 | +21 |
| `get_instance_info` | 604 → 710 | +106 | 127 → 156 | +29 |
| `get_layer_annotation_info` | 905 → 996 | +91 | 196 → 219 | +23 |
| `get_query_info` | 848 → 919 | +71 | 195 → 216 | +21 |
| `get_report_info` | 1,159 → 1,232 | +73 | 248 → 269 | +21 |
| `get_rls_filter_info` | 761 → 840 | +79 | 176 → 201 | +25 |
| `get_role_info` | 675 → 744 | +69 | 156 → 177 | +21 |
| `get_saved_query_info` | 935 → 1,016 | +81 | 224 → 247 | +23 |
| `get_schema` | 876 → 941 | +65 | 183 → 202 | +19 |
| `get_table` | 3,453 → 3,707 | +254 | 798 → 877 | +79 |
| `get_tag_info` | 825 → 892 | +67 | 184 → 205 | +21 |
| `get_task_info` | 888 → 957 | +69 | 200 → 221 | +21 |
| `get_theme_info` | 809 → 880 | +71 | 173 → 194 | +21 |
| `get_user_info` | 805 → 874 | +69 | 184 → 203 | +19 |
| `health_check` | 590 → 590 | +0 | 127 → 127 | +0 |
| `list_annotation_layers` | 2,245 → 2,540 | +295 | 499 → 574 | +75 |
| `list_charts` | 4,635 → 4,946 | +311 | 1,030 → 1,118 | +88 |
| `list_dashboards` | 4,120 → 4,419 | +299 | 918 → 1,000 | +82 |
| `list_databases` | 3,119 → 3,386 | +267 | 694 → 767 | +73 |
| `list_datasets` | 4,195 → 4,486 | +291 | 919 → 1,000 | +81 |
| `list_layer_annotations` | 2,481 → 2,738 | +257 | 569 → 634 | +65 |
| `list_metrics` | 1,527 → 1,708 | +181 | 315 → 368 | +53 |
| `list_queries` | 2,555 → 2,812 | +257 | 577 → 648 | +71 |
| `list_reports` | 3,458 → 3,717 | +259 | 776 → 847 | +71 |
| `list_rls_filters` | 2,225 → 2,496 | +271 | 509 → 587 | +78 |
| `list_roles` | 2,340 → 2,591 | +251 | 541 → 612 | +71 |
| `list_saved_queries` | 2,524 → 2,801 | +277 | 581 → 656 | +75 |
| `list_tags` | 2,671 → 2,918 | +247 | 634 → 705 | +71 |
| `list_tasks` | 2,245 → 2,508 | +263 | 513 → 586 | +73 |
| `list_themes` | 2,563 → 2,818 | +255 | 600 → 671 | +71 |
| `list_users` | 2,512 → 2,763 | +251 | 580 → 650 | +70 |
| `manage_dashboard_certification` | 1,633 → 1,792 | +159 | 359 → 400 | +41 |
| `manage_dashboard_owners` | 1,943 → 2,032 | +89 | 420 → 441 | +21 |
| `manage_dashboard_roles` | 1,693 → 1,780 | +87 | 360 → 381 | +21 |
| `manage_native_filters` | 5,838 → 6,544 | +706 | 1,228 → 1,438 | +210 |
| `open_sql_lab_with_context` | 1,489 → 1,668 | +179 | 319 → 372 | +53 |
| `query_dataset` | 3,320 → 3,562 | +242 | 770 → 840 | +70 |
| `remove_chart_from_dashboard` | 1,030 → 1,125 | +95 | 216 → 239 | +23 |
| `restore_chart` | 892 → 963 | +71 | 195 → 214 | +19 |
| `restore_dashboard` | 753 → 832 | +79 | 157 → 176 | +19 |
| `save_sql_query` | 1,294 → 1,449 | +155 | 285 → 331 | +46 |
| `update_chart` | 104,163 → 53,395 | -50,768 | 25,227 → 12,759 | -12,468 |
| `update_chart_preview` | 102,352 → 52,409 | -49,943 | 24,838 → 12,563 |
-12,275 |
| `update_dashboard` | 3,598 → 3,983 | +385 | 726 → 843 | +117 |
| `update_dataset_metric` | 2,477 → 2,972 | +495 | 554 → 688 | +134 |
</details>
**CI:** Re-read the rollup at `d6e5bafb`: **85 SUCCESS / 10 SKIPPED / 3
NEUTRAL / 0 FAILURE / 0 PENDING**. `test-load-examples`, `testcontainers
(starrocks, 15)`, `playwright-tests (chromium)`, `babel-extract`, and every
other Playwright job all passed. Between `25e553307f` and that head, only
pytest-mark formatting changed; the `tiktoken`/`regex` dependency pins did not
change. This does not establish the cause of the earlier failures.
The test-only follow-up head `90fe26612e` has a new CI run: 3 IN_PROGRESS /
1 PENDING / 28 QUEUED / 3 SKIPPED as of 2026-09-23 15:14 UTC. Those new-head
checks are not yet complete.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
