25 · Compression & Checksums

Gaps instead of ids, vints instead of fixed widths, blocks instead of whole lists. Checksums catch corruption.

How small can this get, and how do we know it is intact? Answered by Gap encoding, bit packing, CRC32.

In this lab: Compress real postings four ways, then flip a byte and watch the checksum catch it.

loading mind map…

building the index…