v1.2.8

What's Changed

• docs: fix hit ratio vs filtered-out ratio in ANN tuning notebook by @thomasht86 in #1352
• feat: accept data plane cert/key content in Vespa() by @thomasht86 in #1353
• add support for MatchPhaseRanking total_max_hits parameter by @mackenziestarr in #1354
• Change pruning age for proton in performance tests by @sebastiannberg in #1356
• ci: sync the vendored metric preset list with the Vespa CLI by @thomasht86 in #1349
• Add retry controls and pass real error status to feed callbacks by @sebastiannberg in #1357
• Make delete_all_docs retry failed chunks and raise instead of silently stopping by @sebastiannberg in #1358
• Split the performance suite into a k6 instance ceiling and a single-process pyvespa lane by @sebastiannberg in #1359
• Let feed_async_iterable really run max_workers requests in flight by @sebastiannberg in #1360
• Size the pyvespa lane by round trip so far runners stop looking like regressions by @sebastiannberg in #1361
• Pass max_concurrent through to the client in query_many and warn above 1024 threads by @sebastiannberg in #1363
• Add a gzip on/off performance test with a 4 KB document in both lanes by @sebastiannberg in #1362
• Keep a performance session under an hour by @sebastiannberg in #1365
• Run the gzip performance test on mTLS only and set its floors by @sebastiannberg in #1364
• Bump urllib3 to 2.8.0 + all deps in pyvespa (2 Mend HIGH CVEs) by @odosk in #1366
• Support listing and revoking data-plane tokens by @ean in #1367

New Contributors

• @mackenziestarr made their first contribution in #1354

Full Changelog: v1.2.7...v1.2.8