Custom parser documentation doesn't specify Read + Seek bounds come from binread::io not std::io

[anieuwland]

I'm trying to define a custom parser, but I'm getting typing errors that make it impossible to do any parsing. Basically, it looks like this:

use std::io::Read;
use std::io::Seek;
use byteorder::{BigEndian, ReadBytesExt};
use binread::*;
// (others left out for brevity, removing imports doesn't solve issue however)

#[derive(Debug)]
#[derive(BinRead)]
struct FlirRawData {
    #[br(pad_before = 2)]
    #[br(little)]
    image_width: u16,
    #[br(little)]
    image_height: u16,
    image_type: u16,
    #[br(pad_before = 26)]
    #[br(parse_with = raw_image_parser)]
    raw_image Vec<u8>,
}

fn raw_thermal_parser<R: Read + Seek>(reader: &mut R, ro: &ReadOptions, _: ()) -> BinResult<Vec<u8>> {
    Ok(Vec::new())
}

Basically, it's like straight from the documentation. Unfortunately this fails to compile with the error that the trait bounds aren't satisfied (as below). If I remove the Read trait from the R type in the function definition, then it complains about Seek not being implemented for R. If I remove both, then it compiles, but then R can't be used for parsing in any way that I can see.

error[E0277]: the trait bound `R: std::io::Read` is not satisfied
   --> src/thermograms/flir.rs:174:23
    |
174 |     #[br(parse_with = raw_image_parser)]
    |                       ^^^^^^^^^^^^^^^^^^ the trait `std::io::Read` is not implemented for `R`

What's going wrong here?

[jam1garner]

Ah ok! I should definitely clarify that in the docs. But Read + Seek there refers to binread::io::{Read, Seek};, which are what allows binread to work in no_std.

If you change your imports to use those, it will work. The interface should be the same, and binread::io::Read is implemented for std::io::Read, but unfortunately since all of std::io is std-only this is a bit of a hack that's kinda required.

Keeping this issue open as a documentation issue though until I (or someone else) gets around to fixing it.

[jam1garner]

Marking this as closed due to #38 making this probably not needed. Reopen if it's a problem for you still.