Struct sentry_core::protocol::debugid::DebugId

[−]

#[repr(C, packed)]
pub struct DebugId { /* private fields */ }

Expand description

Unique identifier for debug information files and their debug information.

This type is analogous to CodeId, except that it identifies a debug file instead of the actual library or executable. One some platforms, a DebugId is an alias for a CodeId but the exact rules around this are complex. On Windows, the identifiers are completely different and refer to separate files.

The string representation must be between 33 and 40 characters long and consist of:

36 character hyphenated hex representation of the UUID field
1-16 character lowercase hex representation of the u32 appendix

The debug identifier is compatible to Google Breakpad. Use DebugId::breakpad to get a breakpad string representation of this debug identifier.

There is one exception to this: for the old PDB 2.0 format the debug identifier consists of only a 32-bit integer + age resulting in a string representation of between 9 and 16 hex characters.

Example

use std::str::FromStr;
use debugid::DebugId;

let id = DebugId::from_str("dfb8e43a-f242-3d73-a453-aeb6a777ef75-a")?;
assert_eq!("dfb8e43a-f242-3d73-a453-aeb6a777ef75-a".to_string(), id.to_string());

In-memory representation

The in-memory representation takes up 32 bytes and can be directly written to storage and mapped back into an object reference.

use std::str::FromStr;
use debugid::DebugId;

let debug_id = DebugId::from_str("dfb8e43a-f242-3d73-a453-aeb6a777ef75-a").unwrap();

let slice = &[debug_id];
let ptr = slice.as_ptr() as *const u8;
let len = std::mem::size_of_val(slice);
let buf: &[u8] = unsafe { std::slice::from_raw_parts(ptr, len) };

let mut new_buf: Vec<u8> = Vec::new();
std::io::copy(&mut std::io::Cursor::new(buf), &mut new_buf).unwrap();

let ptr = new_buf.as_ptr() as *const DebugId;
let new_debug_id = unsafe { &*ptr };

assert_eq!(*new_debug_id, debug_id);

As long the bytes were written using the same major version of this crate you will be able to read it again like this.

Implementations

impl DebugId

pub fn nil() -> DebugId

Constructs an empty debug identifier, containing only zeros.

pub fn from_uuid(uuid: Uuid) -> DebugId

Constructs a DebugId from its uuid.

pub fn from_parts(uuid: Uuid, appendix: u32) -> DebugId

Constructs a DebugId from its uuid and appendix parts.

pub fn from_guid_age(
guid: &[u8],
age: u32
) -> Result<DebugId, ParseDebugIdError>

Constructs a DebugId from a Microsoft little-endian GUID and age.

pub fn from_pdb20(timestamp: u32, age: u32) -> DebugId

Constructs a DebugId from a PDB 2.0 timestamp and age.

pub fn from_breakpad(string: &str) -> Result<DebugId, ParseDebugIdError>

Parses a breakpad identifier from a string.

pub fn uuid(&self) -> Uuid

Returns the UUID part of the code module’s debug_identifier.

If this is a debug identifier for the PDB 2.0 format an invalid UUID is returned where only the first 4 bytes are filled in and the remainder of the bytes are 0. This means the UUID has variant [uuid::Variant::NCS] and an unknown version, Uuid::get_version will return None, which is not a valid UUID.

This may seem odd however does seem reasonable:

Every DebugId can be represented as Uuid and will still mostly look reasonable e.g. in comparisons etc.
The PDB 2.0 format is very old and very unlikely to appear practically.