serialize: add json::{Integer,Floating} to parse large integers properly #16201
Conversation
|
I really like this, it is my biggest frustration with many C Json libraries that they do not handle integers precisely when the underlying language is perfectly able to. |
|
There is another problem with Json-decoded integers: if you try to read an unsigned integer, it'll read an f64 and cast this to a uint, resulting in wraparound. I think the decoder ought to check for negative numbers and signal an error. Maybe this is a job for another PR though? |
|
@apoelstra: Right now this patch can properly deserialize i64s, although it doesn't check for overflow. I should probably add that. I could also add support for u64s, which would let us get that last bit of size. |
|
@erickt Full-range support for u64's would be awesome, though personally I would only use it for the side-effect of rejecting negative numbers :) |
|
This is ready again for review. I added support for |
|
This looks great -- I like your use of |
| macro_rules! read_primitive { | ||
| ($name:ident, $ty:ty) => { | ||
| fn $name(&mut self) -> DecodeResult<$ty> { | ||
| println!("$name {}", self.stack); |
|
Sorry for the delay @erickt, but looks good to me! r=me with two minor things addressed. |
This patch allows json to deserialize integers larger than 2^53 without losing precision. It does this by first keeping the integer portion of a number as a `i64`, and only casting it over to a `f64` if we have a decimal or exponent.
This patch allows json to deserialize integers larger than 2^53 without losing precision. It does this by first keeping the integer portion of a number as a
i64, and only casting it over to af64if we have a decimal or exponent.