# What changed in @orama/core

> 8 changes in @orama/core 1 from its official release notes, upgrade guides and advisories. Each one cited. Sources checked 2026-10-19. HTML: https://unstale.sh/npm/@orama/core/

To see which of these apply to a repository, run `npx @unstale/cli init` in it.

## @orama/core 1

### Performance

- Scans are bounded to page slots and read in 64 KiB chunks, so lookups of absent documents no longer scan the whole page file. (1.2.39) — [release notes](https://github.com/oramasearch/oramacore/releases/tag/v1.2.39)
- Avoids taking a write lock on the index. (1.2.38) — [release notes](https://github.com/oramasearch/oramacore/releases/tag/v1.2.38)

### Behavior and default changes

- The informational document count saturates instead of blocking inserts. (1.2.39) — [release notes](https://github.com/oramasearch/oramacore/releases/tag/v1.2.39)
- Zebo is compacted at startup. (1.2.38) — [release notes](https://github.com/oramasearch/oramacore/releases/tag/v1.2.38)

### Fixes and other changes

- Entries beyond document_limit written by older versions remain findable, and rotation respects each page's own limit. (1.2.39) — [release notes](https://github.com/oramasearch/oramacore/releases/tag/v1.2.39)
- A page with inconsistent offsets is no longer written to; new documents are sent to a new page. (1.2.39) — [release notes](https://github.com/oramasearch/oramacore/releases/tag/v1.2.39)
- Index entries pointing outside the file are skipped. (1.2.39) — [release notes](https://github.com/oramasearch/oramacore/releases/tag/v1.2.39)
- Pages rotate on byte size and write offsets are checked, preventing a page from exceeding the u32 offset range and wrapping over its own index. (1.2.39) — [release notes](https://github.com/oramasearch/oramacore/releases/tag/v1.2.39)