Module num 

Redis’s own integer and double text, shared with the string type.

This module lives in yo-common because the codec is not the only thing that needs it: whether a string is stored int encoded is decided by the same string2ll rules that decide whether a bulk length parses. Re-exported here so that yo_resp::num keeps meaning what it meant. Numbers to text and back, written by hand.

Every reply that carries a length or an integer goes through here, which is every reply, so this is as hot as anything in the codec. The formatting machinery in core::fmt would produce the same bytes and would be several times slower for the two or three digits a bulk header usually needs, so it is not used for integers.

Parsing is deliberately strict and matches Redis’s string2ll byte for byte, including its refusal of leading zeros and of a leading +. The protocol’s own lengths are parsed with it, so a stricter or looser reading here is a real difference in what the two servers accept.

Constants

DIGITS_MAX

A buffer big enough for the digits of any i64 or u64, sign included.

DOUBLE_MAX

Room for the longest thing write_double or write_g17 can write.

Functions

i64_digits

The decimal digits of n, written into the back of buf.

i64_len

The number of bytes push_i64 would append.

parse_f64

Redis’s getLongDoubleFromObject, as far as the difference is observable.

parse_i64

Parses a signed decimal integer the way Redis’s string2ll does.

push_double

Appends a double the way Redis 8 writes one.

push_fixed4

Appends a distance the way the geo commands write one, which is four digits after the point and no exponent ever.

push_g17

The same digits write_g17 would write, appended.

push_human

Appends a double the way INCRBYFLOAT and HINCRBYFLOAT write one, which is not the way everything else does.

push_i64

Appends the decimal digits of n, with a minus sign if it needs one.

push_u64

Appends the decimal digits of n.

u64_digits

The decimal digits of n, written into the back of buf.

u64_len

How many digits n has.

write_double

Writes a double into a fixed buffer, byte for byte what push_double would append.

write_g17

Writes a double the way C’s %.17g writes one, which is what AROP replies with.