Oleksiy Syvokon f0287d69e5 ep: Compress EP requests (#48793) ...

Benchmark results on 3,000 captured requests:

| Scenario          | Mean   | Savings | p95    | Savings |
|:------------------|-------:|--------:|-------:|--------:|
| Uncompressed      | 12,275 |       — | 35,896 |       — |
| zstd (no dict)    |  2,784 |  77.3%  |  7,193 |  80.0%  |
| zstd + global dict|  2,544 |  79.3%  |  6,862 |  80.9%  |
| brotli (no dict)  |  2,650 |  78.4%  |  6,857 |  80.9%  |

A trained zstd dictionary adds only 2% over plain zstd, which is not
worth added complexity.

Brotli compresses comparably but is 4 times slower.

Release Notes:

- N/A

2026-02-09 18:02:56 +02:00

..

license_examples

Reorganize edit prediction code and remove old experiments (#44187)

2025-12-04 15:56:57 -08:00

license_patterns

Reorganize edit prediction code and remove old experiments (#44187)

2025-12-04 15:56:57 -08:00

src

ep: Compress EP requests (#48793)

2026-02-09 18:02:56 +02:00

Cargo.toml

ep: Compress EP requests (#48793)

2026-02-09 18:02:56 +02:00

LICENSE-GPL

Rename remaining mentions of "inline completion" to "edit prediction" (#35512)

2025-08-04 16:22:18 +00:00