|0.1.0||Aug 29, 2020|
A Better Varint encoding that preserves
We use variable length integers.
[snip: how traditional VLQ works]
This was a mistake. If you need variable length integers, don't do them this way. Instead, do them like this where the first byte tells you the magnitude of the integer.
This is very important for efficiency in parsing. The other thing is that you can actually compare two integers using
memcmpwithout having to decode them.
A Better Varint
Based on D. Richard Hipp.'s "A Better Varint" idea. See https://youtu.be/gpxnbly9bz4?t=2386.
Slightly changed so leading 0xff is reserved for larger integers.