This coding guide is inspired from the Uber Go style guide at https://github.com/uber-go/guide/blob/master/style.md, with modifications specific to Cockroach Labs.

Introduction

These guidelines are the conventions that govern our code. These conventions cover far more than just source file formatting—gofmt handles that for us.

The goal of this guide is to manage this complexity by describing in detail the Dos and Don'ts of writing Go code at Cockroach Labs. These rules exist to keep the code base manageable while still allowing engineers to use Go language features productively.

This documents idiomatic conventions in Go code that we follow at CRL. A lot of these are general guidelines for Go, while others extend upon external resources:

1. Effective Go

2. The Go common mistakes guide

All code should be error-free when run through golint and go vet. We recommend setting up your editor to:

• Run crlfmt on save

• Run make lintshort to check for errors

You can find information in editor support for Go tools here: golang/go/wiki/IDEsAndTextEditorPlugins

Guidelines

Pointers to Interfaces

You almost never need a pointer to an interface. You should be passing interfaces as values—the underlying data can still be a pointer.

An interface is two fields:

1. A pointer to some type-specific information. You can think of this as "type."

2. Data pointer. If the data stored is a pointer, it’s stored directly. If the data stored is a value, then a pointer to the value is stored.

If you want interface methods to modify the underlying data, you must use a pointer.

Receivers and Interfaces

Methods with value receivers can be called on pointers as well as values.

For example,

Similarly, an interface can be satisfied by a pointer, even if the method has a value receiver.

Effective Go has a good write up on Pointers vs. Values.

Zero-value Mutexes are Valid

The zero-value of sync.Mutex and sync.RWMutex is valid, so you almost never need a pointer to a mutex.

If you use a struct by pointer, then the mutex can be a non-pointer field or, preferably, embedded directly into the struct.

Copy Slices and Maps when necessary

Slices and maps contain pointers to the underlying data so be conscious of ownership when passing them as parameters or returning them as values.

On the other hand, do not unnecessarily copy slices - heap allocations are expensive and we want to avoid them when possible.

Receiving Slices and Maps

Keep in mind that users can modify a map or slice you received as an argument if you store a reference to it.

If your API captures a slice by reference, document it in the function header.

Returning Slices and Maps

Similarly, be wary of user modifications to maps or slices exposing internal state.

Your exported API should refrain from returning internal state that otherwise has accessors, e.g. when it is hidden behind a mutex.

For "internal" functions there is more flexibility, although the function name should reflect this (see e.g. log.closeFileLocked in pkg/util/log).

Defer to Clean Up

Use defer to clean up resources such as files and locks.

Defer has an extremely small overhead and should be avoided only if you can prove that your function execution time is in the order of nanoseconds. The readability win of using defers is worth the minuscule cost of using them. This is especially true for larger methods that have more than simple memory accesses, where the other computations are more significant than the defer

Sometimes, you may wish to Lock a mutex, do some work, and the Unlock it before the end of the function has been reached. If the work being done in this critical section has the potential to panic, this could end up leaving the mutex locked if the executing code is wrapped in a panic handler, leading to deadlocks.

To protect against this, execute the code in a separate function so that defer can still be used. This way, even if a panic occurs in the critical section, the mutex will be released.

Channel Size is One or None

Channels should usually have a size of one or be unbuffered. By default, channels are unbuffered and have a size of zero. Any other size must be subject to a high level of scrutiny. Consider how the size is determined, what prevents the channel from filling up under load and blocking writers, and what happens when this occurs.

Start Enums at One

The standard way of introducing enumerations in Go is to declare a custom type and a const group with iota. Since variables have a 0 default value, you should usually start your enums on a non-zero value.

There are cases where using the zero value makes sense, for example when the zero value case is the desirable default behavior.

Error handling

The following sub-section are an excerpt from https://cockroachlabs.atlassian.net/wiki/pages/resumedraft.action?draftId=1676804870 . Read the full page for more background and context.

Performance

Performance-specific guidelines apply only to the hot path.

Be mindful that strconv is faster than fmt

When converting primitives to/from strings, strconv is faster than fmt.

Avoid string-to-byte conversion

Do not create byte slices from a fixed string repeatedly: it causes a heap allocation and a copy. Instead, perform the conversion once and capture the result.

Style

Code comments

See https://wiki.crdb.io/wiki/spaces/CRDB/pages/181371303/Go+Golang+coding+guidelines

Line Length

Format your code assuming it will be read in a window 100 columns wide. Wrap code at 100 characters and comments at 80 unless doing so makes the code less legible. These values assume tab width is 2 characters.

Wrapping Function Signatures

When wrapping function signatures that do not fit on one line, put the name, arguments, and return types on separate lines, with the closing ) at the same indentation as func (this helps visually separate the indented arguments from the indented function body). Example:

If the arguments list is too long to fit on a single line, switch to one argument per line:

If the return types need to be wrapped, use the same rules:

Exception when omitting repeated types for consecutive arguments: short and related arguments (e.g. `start, end int64`) should either go on the same line or the type should be repeated on each line -- no argument should appear by itself on a line with no type  (confusing and brittle when edited).

Group Similar Declarations

Go supports grouping similar declarations.

This also applies to constants, variables, and type declarations.

Only group related declarations. Do not group declarations that are unrelated.

Groups are not limited in where they can be used. For example, you can use them inside of functions.

Import Group Ordering

There should be two import groups:

• Standard library

• Everything else

This is the grouping applied by goimports by default.

Package Names

When creating and naming packages, refer to the separate page https://cockroachlabs.atlassian.net/wiki/spaces/CRDB/pages/1849657053 .

Function Names

We follow the Go community's convention of using MixedCaps for function names. An exception is made for test functions, which may contain underscores for the purpose of grouping related test cases, e.g., TestMyFunction_WhatIsBeingTested.

Import Aliasing

Import aliasing must be used if the package name does not match the last element of the import path.

In all other scenarios, import aliases should be avoided unless there is a direct conflict between imports.

Function Grouping and Ordering

• Functions should be sorted in rough call order.

• Functions in a file should be grouped by receiver.

Therefore, exported functions should appear first in a file, after struct, const, var definitions.

A newXYZ()/NewXYZ() may appear after the type is defined, but before the rest of the methods on the receiver.

Since functions are grouped by receiver, plain utility functions should appear towards the end of the file.

Reduce Nesting

Code should reduce nesting where possible by handling error cases/special conditions first and returning early or continuing the loop. Reduce the amount of code that is nested multiple levels.

Unnecessary Else

Avoid "else" clauses if they're not needed.

If a variable is set in both branches of an if, it can be replaced with a single if:

A return statement inside of an if block means that you won't need an "else" clause.

Top-level Variable Declarations

At the top level, use the standard var keyword. Do not specify the type, unless it is not the same type as the expression.

Specify the type if the type of the expression does not match the desired type exactly.

Use Field Names to initialize Structs

You should almost always specify field names when initializing structs. This is now enforced by go vet.

Exception: Field names may be omitted in test tables when there are 3 or fewer fields.

Local Variable Declarations

Short variable declarations (:=) should be used if a variable is being set to some value explicitly.

However, there are cases where the default value is clearer when the var keyword is use. Declaring Empty Slices, for example.

nil is a valid slice

nil is a valid slice of length 0. This means that,

You should not return a slice of length zero explicitly. Return nil instead.

To check if a slice is empty, always use len(s) == 0. Do not check for nil.

The zero value (a slice declared with var) is usable immediately without make().

Reduce Scope of Variables

Where possible, reduce scope of variables. Do not reduce the scope if it conflicts with Reduce Nesting.

If you need a result of a function call outside of the if, then you should not try to reduce the scope.

Avoid Naked Parameters

Naked parameters in function calls can hurt readability. Add C-style comments (/* ... */) for parameter names when their meaning is not obvious.

Better yet, replace naked bool types with custom types for more readable and type-safe code. This allows more than just two states (true/false) for that parameter in the future.

Note: For `bool` constants, like for all literals, the comment should indicate the name of the parameter and does not depend on the argument value. *Do not* put a bang in the comment when commenting the `false` constant. Also, do not adapt the comment to negate the name of the parameter. For example:

Try to avoid bool parameters

bool arguments to functions are often dubious things, as they hint to code that essentially reads like:

This is not always the case. However, in cases where that is a fair assessment of the situation, consider whether the doSomething function should exist at all.

In cases where the `bool` in question, along with other arguments, acts as a "knob" to the function consider replacing it with some type of "configuration" struct (for examples, see [Dave Cheney's treatment of the topic](https://dave.cheney.net/2014/10/17/functional-options-for-friendly-apis)). In situations where there's a single bool param or the situation is less clear-cut, consider replacing the `bool` in question with an enum. For example:

is better than:

fmt Verbs

Prefer the most specific verb for your use. In other words, prefer to avoid %v when possible. However, %v is to be used when formatting bindings which might be nil and which do not already handle nil formatting. Notably, nil errors formatted as %s will render as %!s(<nil>) while nil errors formatted as %v will render as <nil>. Therefore, prefer %v when formatting errors which are not known to be non-nil.

Use Raw String Literals to Avoid Escaping

Go supports raw string literals, which can span multiple lines and include quotes. Use these to avoid hand-escaped strings which are much harder to read.

Initializing Struct References

Use &T{} instead of new(T) when initializing struct references so that it is consistent with the struct initialization.

Format Strings outside Printf

If you declare format strings for Printf-style functions outside a string literal, make them const values.

This helps go vet perform static analysis of the format string.

Naming Printf-style Functions

When you declare a Printf-style function, make sure that go vet can detect it and check the format string.

This means that you should use pre-defined Printf-style function names if possible. go vet will check these by default. See Printf family for more information.

If using the pre-defined names is not an option, end the name you choose with f: Wrapf, not Wrap. go vet can be asked to check specific Printf-style names but they must end with f.

See also go vet: Printf family check.

Go implementation of SQL

When defining the result column labels of statements or virtual tables, apply the following general principles:

• the casing of column labels must be consistent.

• the same concept in different statements should be named using the same word.
  For example, `variable` and `value` corresponds between SHOW ALL CLUSTER SETTINGS and SHOW SESSION ALL.

• the labels must be usable by a surrounding query without mandating quotes.
  For example, `start_key` instead of `"Start Key"`. Also, avoid SQL keywords which would also need to be quoted: `table_name` instead of `"table"`, `index_name` instead of `"index"`, etc.
  Never name a column `"user"`! The risk of mistake with the syntax special form `user` is too high. Use `user_name` instead.

• use underscores to separate words, for consistency with `information_schema`.
  For example, `start_key` instead of `startkey`

• avoid abbreviations unless most uses of the word in spoken English use the abbreviation already.
  For example, `id` for "identifier" is OK, but `loc` for "location" is not.

• for columns akin to "primary keys" (that identify the object being described by the remaining columns), words whose meaning is highly overloaded must be disambiguated in the label.
  For example, `zone_id`, `job_id`, `table_id` instead of just `id` everywhere. `table_name`, `column_name`, etc instead of `name` everywhere.

• for database objects that can be referred to by two or more types of handles (e.g. by-ID or by-name), the column label must disambiguate which type of handle is being reported. This makes it possible to later report the other handles in additional columns.
  For example, `table_name` instead of `table`, so that `table_id` can be added later.

• labels that reproduce data already present in an `information_schema` table or a prior `SHOW` statement should use the same labels as the `information_schema` table or `SHOW` statement, provided it matches the principles above already.

Patterns

Data-driven tests

We provide an alternative to table-driven tests (see next section) in CockroachDB via the datadriven test package.

Test Tables

Use table-driven tests with subtests to avoid duplicating code when the core test logic is repetitive.

Test tables make it easier to add context to error messages, reduce duplicate logic, and add new test cases.

We follow the convention that the slice of structs is referred to as tests and each test case tt. Further, we encourage explicating the input and output values for each test case with give and want prefixes.

Functional Options

Functional options is a pattern in which you declare an opaque Option type that records information in some internal struct. You accept a variadic number of these options and act upon the full information recorded by the options on the internal struct.

Use this pattern for optional arguments in constructors and other public APIs that you foresee needing to expand, especially if you already have three or more arguments on those functions.

See also about functional options:

• Self-referential functions and the design of options

• Functional options for friendly APIs