4 ms·
If I'm going to do that I think I'd use Bitcoins BASE58 which avoids letters that could be confused for each other. The number of times I see an O and 0 and won
by giancarlostoro 2y ago
If I'm going to do that I think I'd use Bitcoins BASE58 which avoids letters that could be confused for each other. The number of times I see an O and 0 and wonder which is which, because the font does not make it clear really annoys me.
Edit:
Other honorable mentions: ObjectID's as used by MongoDB which contain the creation timestamp. Also Discord's snowflakes (inspired by Twitter's iirc), which also contain the creation timestamp.
- kibwen 2y agoIf you want random IDs to be human-readable (and human-communicatable), I'd just recommend base 32 or even base16. You don't actually save that many bytes from base58 or base64 when it comes to short IDs. Case in point, the parent poster's base64 ID is 14 characters long. When encoded as base32 that's still only 17 characters (or 19 in base16), and now you have completely gotten rid of all notion of casing, which is annoying to communicate verbally.
- giancarlostoro 2y agoI think the more I think of it, the more I favor the Discord snowflake ID's because they're just integers, and they can be generated on the fly. I think new messages generated them on the client if I'm not mistaken.