# the bendncode format specification ## definitions - leb128: store arbitrary size integers as multiples of 7 bits, with 8th bit as continuation - sleb128: similar but you write the sign bit at the end ## decoding steps as this is a self descriptive format, you need not know what it must become to read what it is. the first step of any decode, is to read the tag out. none of the possible variants have a closing tag, merely an opening tag. - `\x0` (False) and `\x1` (True) map straight to a boolean - `i` (Int) uses the `sleb128` encoding as defined - `u` (Uint) uses the `leb128` encoding - `f` (Float) simply encodes 4 bytes - `d` (Double) encodes 8 bytes - `z` (None) means none, and `Some` means read the next tag - `l` (List) is encoded as `leb128(len)`, `tagged contents`, and are not necessarily homogenous. - `s` (String) is encoded as `leb128(len)`, followed by `len` bytes. - `m` (Map) is encoded as `leb128(len)`, `kₙ`, `vₙ`, .. ### enum tags - all enums are, immediately after their tag, encoded with their index(`leb128`) - `v`, for newtype variant, is encoded as simply the actual type (which is tagged). - `x`, unit variant, has no further steps. - `y`, struct variant, is encoded identically to `Map`. - `n`, tuple variant, is encoded identically to `List`. ## flowchart