villebro commented on PR #44508: URL: https://github.com/apache/superset/pull/44508#issuecomment-5778908207
Pushed b950a5fbe5 on top of Amin's calibration fix (d784db26e7). Two things his fix didn't cover: 1. **Stale "100k"/"100 KB" references it missed**: the actual user-facing `ValueError` in `validate_embedded_dimensions_page_size` still said "~100k by default" (now "~50k"), and `string_clip_chars`'s docstring still referenced "the default 100 KB budget" (now 50 KB, plus its inline `4` divisor is now a named `_CLIP_BUDGET_DIVISOR` constant). 2. **`EMBEDDED_DIMENSIONS_MAX_PAGE_SIZE` was still 8**, which was only ever safe under the *old* 100,000-byte default. Measured directly against this repo's own `MetricInfo`/`DimensionInfo` schemas (40 dims/metric, matching the `large_metric_catalog` fixture): page_size=8 serializes to 54,558 bytes with no descriptions and 77,014 with short descriptions — both already over the new 50,000 default. Since `list_metrics` isn't in the response-size guard's truncation-eligible tool categories, an oversized response there hard-errors rather than degrading, so this wasn't just a documentation gap. Lowered the cap to 4 (38,555 bytes / ~22% margin in the with-descriptions case) and updated every embedded-dimensions boundary test accordingly. Also addressed Bito's 3 non-blocking suggestions from the earlier bot review: folded the duplicate `test_small_budget_scales_string_clip_to_fit`/`test_truncates_large_string_fields` tests, named the clip-budget divisor constant, and changed `format_size_limit_error`'s sentinel check from `>=` to `==` since `UNMEASURABLE_RESPONSE_BYTES` is the only failure value. -- 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]
