Peter Dillinger
Peter Dillinger
The thread of work including #12750 and #12959 is good but introduces a potential efficiency issue in backups. In the simplest case, suppose we ingest SSTs from one CF to...
In some recent dev work I discovered that breaking the behavior this test is supposed to check for (details in 6652c94) is no longer caught by the test. And I...
Summary: WIP TODO: Setting things through "table_factory" is broken TODO: Deny SetOptions on block cache options (dubious and not yet made safe with shared_ptr handling) Fixes #10079 Test Plan: added...
Summary: I was going to use this in some code I was working on but ended up not needing it. But it's useful nonetheless and I'm using it in a...
I believe that if a compaction output is somehow corrupted to have extra data after the actual RocksDB written data, we don't detect that until the DB is re-opened and...
If you use `allow_ingest_behind` with the intention of later stitching together a sequence of updates on top of some base data, that currently doesn't respect Delete updates because it might...
This change builds on #14027 and #13965 to complete migration of the HyperClockCache implementation to using the hygienic BitFields API. No semantic change in the implementation details is intended, just...
See https://github.com/corsix/fast-crc32/ though we probably need to tune differently than 4k block size for checksum because our compressed blocks will often be smaller than 4KB, and very often smaller than...
This function (`ZSTD_createDDict_byReference`) is used by RocksDB, and long has been. It is essentially required for digested dictionary support so that we can manage raw bytes ourselves without memory duplication....