Closed khklau closed 8 months ago
First attempt will use rs-snowflake. To take full advantage of the snowflake algorithm we should be feeding proper machine and node identifiers to the generator, but for the MVP we're going to just hard code these IDs.
Using Snowflake is good enough for now. Closing.
Apache provides an identifier for the client connection, but not for the client request. Initially it seems like using the request_rec address would suffice, but that address is likely to be reused at some point since it's probably stored in an APR memory pool.
So we'll have to generate our own identifier. A 128-bit UUID is probably overkill. The 64-bit snowflake by Twitter might be a good candidate.