Skip to main content

Coding Style

The Enarx project follows the Rust coding guidelines.

Any further recommendation we issue should be due to the fact the Rust project has no official guideline on a specific aspect.

If you find one of our coding recommendations to be in contradiction with the Rust coding guidelines, the latter should prevail.

In any case, we will align to new Rust guidelines when published.


In keeping with our general policy on Rust guidelines outlined above, we expect all code submitted to the repository to conform to the Rust formatting guidelines. All Enarx repositories using Rust enforce this requirement as a gating test for new pull requests.

Fortunately, this is an easy requirement to satisfy using rustfmt, Rust's official formatting tool. If not already installed, you can add it using rustup component add rustfmt. You can run it on any crate by running cargo fmt in a crate's root (the same directory that contains Cargo.toml).


Enarx uses semantic versioning.


Comments should be complete sentences with proper grammar.

They should describe both what the code is doing and why it is doing it.

Variable names

The Enarx project prefers the use of descriptive variable names that give an idea of what that variable is being used for. We do not enforce specific guidelines beyond this preference. For example, we will frown upon extremely short C-style names (think b with regards to a boolean variable).

Inline Assembly

Inline assembly is necessary in this project. However, inline assembly statements are also by nature difficult to comprehend. Enforcing a strict, consistent style on these statements will hopefully make them easier to understand in spite of their difficulty.

  1. Every asm!() statement MUST have a Rust comment before the statement (or its surrounding unsafe block) explaining what it does.

  2. Inline assembly MUST format each assembly instruction as a separate string. If more than one instruction is used, each instruction string MUST appear on a new line.

  3. Each assembly instruction string MAY have a Rust comment at the end of the line explaing what it does. These comments SHOULD be vertically aligned if practical.

  4. Assembly comments MUST NOT be used.

  5. Instruction operands SHOULD be vertically aligned if practical.

  6. Long asm!() statements containing many instructions SHOULD be avoided. Find a way to break them up into smaller chunks if at all possible.

  7. In longer asm!() statements, instruction strings SHOULD be logically grouped with an empty line between them. Each group MAY have a Rust comment explaining what the group does.

  8. Where practical, asm!() statements SHOULD be encapsulated by their own function.

  9. An empty line SHOULD be placed between instruction strings and arguments in a multi-line asm!() statement.


// This is a single line asm.
asm!("ud2", options(noreturn));

// This is multi-line asm.
// Instruction operands and comments are aligned vertically.
// Move inputs into the correct registers
"mov r10, {0}",
"mov r11, {1}",

// Perform summation
"add r10, r11", // Add the inputs
"add r10, {CONSTANT}", // Add the constant

CONSTANT = const 42usize,
in(reg) 7,
in(reg) 12,
out("r10") sum,
out("r11") _,


One nice thing that cargo fmt does is wrap method invocations to improve readability. However, in some cases this wrapping is more harmful than beneficial.

  1. Whenever wrapping occurs, consider using an intermediate variable. A good rule of thumb is that if a single intermediate variable causes no wrapping to occur, it is probably better to use the intermediate variable. This is not a hard and fast rule. Try to use good judgement.

  2. Wrapping MUST NOT occur in any position that can cause a branch. Examples of this include the boolean value of an if statement or while loop. This makes understanding the flow of the program more difficult. Use an intermediate variable to separate the logic producing the boolean from the control flow statements. For example:

   let dangerous = object

if dangerous {

Use Declarations

use-statements are divided into following groups:

  1. Reexports from the same crate.
  2. Reexports from the core, alloc or std.
  3. Reexports from other crates.
  4. Imports from the same crate.
  5. Imports from the core, alloc or std.
  6. Imports from other crates.

Each group is separated by an empty line.

The motivation behind all this is that this makes cargo fmt to sort each group separately, but the groups are still kept in the same relative order.

Import paths

  1. Default to use super::...;. This reduces code churn and aligns with Rust visibility rules.
  2. If the import line gets long, consider using chaining. This is particularly true if the type is used in parent or grandparent crates already.
  3. If the import line remains long or exists in a parallel hierarchy, consider using crate instead.

As a general guideline, prefer super::super::Foo syntax when importing something defined within a direct ancestor up to 2 levels up and crate::a::Foo otherwise.


pub use foo::{Bar, Baz}; // Group #1

use crate::Bat; // Group #4
use super::super::Bag; // Group #4

use core::convert::{TryFrom, TryInto}; // Group #5
use core::num::NonZeroU8; // Group #5

use noted::noted; // Group #6
use primordial::Page; // Group #6

mod tests {
use super::*; // Group #4