Chaser Zaks
Implementation: https://gist.github.com/FelixWolf/8ef1ced3cd1aa37c5dc803d7a79e43d8
(Contributor agreement
should
cover this, I think?)
I requested this on the Jira back when it was around, but it was closed due to endianness issues on various machines.
I've written a implementation that packs the float into base64, and visa versa, regardless of endianness.
It currently encodes in big-endian, which llIntegerToBase64 and llBase64ToInteger do.
It would be really useful if we need to preserve floating point values in a short string.
Benefits of this over doing (string) for floats is that:
  1. Extremely precise floats don't end up like 1.258532194328154328
  2. LSL cuts off floating point precision after the 5th precision (EG: 0.12345 instead of 0.1234567890)