List of all available rules.
- Description of available rules
- add-constant
- argument-limit
- atomic
- banned-characters
- bare-return
- blank-imports
- bool-literal-in-expr
- call-to-gc
- cognitive-complexity
- comment-spacings
- comments-density
- confusing-naming
- confusing-results
- constant-logical-expr
- context-as-argument
- context-keys-type
- cyclomatic
- datarace
- deep-exit
- defer
- dot-imports
- duplicated-imports
- early-return
- empty-block
- empty-lines
- enforce-map-style
- enforce-repeated-arg-type-style
- enforce-slice-style
- error-naming
- error-return
- error-strings
- errorf
- exported
- file-header
- flag-parameter
- function-length
- function-result-limit
- get-return
- identical-branches
- if-return
- import-alias-naming
- import-shadowing
- imports-blocklist
- increment-decrement
- indent-error-flow
- line-length-limit
- max-control-nesting
- max-public-structs
- modifies-parameter
- modifies-value-receiver
- nested-structs
- optimize-operands-order
- package-comments
- range-val-address
- range-val-in-closure
- range
- receiver-naming
- redefines-builtin-id
- redundant-import-alias
- string-format
- string-of-int
- struct-tag
- superfluous-else
- time-equal
- time-naming
- unchecked-type-assertion
- unconditional-recursion
- unexported-naming
- unexported-return
- unhandled-error
- unnecessary-stmt
- unreachable-code
- unused-parameter
- unused-receiver
- use-any
- useless-break
- var-declaration
- var-naming
- waitgroup-by-value
Description: Suggests using constant for magic numbers and string literals.
Configuration:
maxLitCount
: (string) maximum number of instances of a string literal that are tolerated before warn.allowStr
: (string) comma-separated list of allowed string literalsallowInts
: (string) comma-separated list of allowed integersallowFloats
: (string) comma-separated list of allowed floatsignoreFuncs
: (string) comma-separated list of function names regexp patterns to exclude
Example:
[rule.add-constant]
arguments = [{ maxLitCount = "3", allowStrs = "\"\"", allowInts = "0,1,2", allowFloats = "0.0,0.,1.0,1.,2.0,2.", ignoreFuncs = "os\\.*,fmt\\.Println,make" }]
Description: Warns when a function receives more parameters than the maximum set by the rule's configuration. Enforcing a maximum number of parameters helps to keep the code readable and maintainable.
Configuration: (int) the maximum number of parameters allowed per function.
Example:
[rule.argument-limit]
arguments = [4]
Description: Check for commonly mistaken usages of the sync/atomic
package
Configuration: N/A
Description: Checks given banned characters in identifiers(func, var, const). Comments are not checked.
Configuration: This rule requires a slice of strings, the characters to ban.
Example:
[rule.banned-characters]
arguments = ["Ω","Σ","σ"]
Description: Warns on bare (a.k.a. naked) returns
Configuration: N/A
Description: Blank import should be only in a main or test package, or have a comment justifying it.
Configuration: N/A
Description: Using Boolean literals (true
, false
) in logic expressions may make the code less readable. This rule suggests removing Boolean literals from logic expressions.
Configuration: N/A
Description: Explicitly invoking the garbage collector is, except for specific uses in benchmarking, very dubious.
The garbage collector can be configured through environment variables as described here.
Configuration: N/A
Description: Cognitive complexity is a measure of how hard code is to understand. While cyclomatic complexity is good to measure "testability" of the code, cognitive complexity aims to provide a more precise measure of the difficulty of understanding the code. Enforcing a maximum complexity per function helps to keep code readable and maintainable.
Configuration: (int) the maximum function complexity
Example:
[rule.cognitive-complexity]
arguments = [7]
Description: Spots comments of the form:
//This is a malformed comment: no space between // and the start of the sentence
Configuration: ([]string) list of exceptions. For example, to accept comments of the form
//mypragma: activate something
//+optional
You need to add both "mypragma:"
and "+optional"
in the configuration
Example:
[rule.comment-spacings]
arguments = ["mypragma:", "+optional"]
Description: Spots files not respecting a minimum value for the comments lines density metric = comment lines / (lines of code + comment lines) * 100
Configuration: (int) the minimum expected comments lines density.
Example:
[rule.comments-density]
arguments = [15]
Description: Methods or fields of struct
that have names different only by capitalization could be confusing.
Configuration: N/A
Description: Function or methods that return multiple, no named, values of the same type could induce error.
Configuration: N/A
Description: The rule spots logical expressions that evaluate always to the same value.
Configuration: N/A
Description: By convention, context.Context
should be the first parameter of a function. This rule spots function declarations that do not follow the convention.
Configuration:
allowTypesBefore
: (string) comma-separated list of types that may be before 'context.Context'
Example:
[rule.context-as-argument]
arguments = [{allowTypesBefore = "*testing.T,*github.com/user/repo/testing.Harness"}]
Description: Basic types should not be used as a key in context.WithValue
.
Configuration: N/A
Description: Cyclomatic complexity is a measure of code complexity. Enforcing a maximum complexity per function helps to keep code readable and maintainable.
Configuration: (int) the maximum function complexity
Example:
[rule.cyclomatic]
arguments = [3]
Description: This rule spots potential dataraces caused by go-routines capturing (by-reference) particular identifiers of the function from which go-routines are created. The rule is able to spot two of such cases: go-routines capturing named return values, and capturing for-range
values.
Configuration: N/A
Description: Packages exposing functions that can stop program execution by exiting are hard to reuse. This rule looks for program exits in functions other than main()
or init()
.
Configuration: N/A
Description: This rule warns on some common mistakes when using defer
statement. It currently alerts on the following situations:
name | description |
---|---|
call-chain | even if deferring call-chains of the form foo()() is valid, it does not helps code understanding (only the last call is deferred) |
loop | deferring inside loops can be misleading (deferred functions are not executed at the end of the loop iteration but of the current function) and it could lead to exhausting the execution stack |
method-call | deferring a call to a method can lead to subtle bugs if the method does not have a pointer receiver |
recover | calling recover outside a deferred function has no effect |
immediate-recover | calling recover at the time a defer is registered, rather than as part of the deferred callback. e.g. defer recover() or equivalent. |
return | returning values form a deferred function has no effect |
These gotchas are described here
Configuration: by default all warnings are enabled but it is possible selectively enable them through configuration. For example to enable only call-chain
and loop
:
[rule.defer]
arguments = [["call-chain","loop"]]
Description: Importing with .
makes the programs much harder to understand because it is unclear whether names belong to the current package or to an imported package.
More information here
Configuration:
allowedPackages
: (list of strings) comma-separated list of allowed dot import packages
Example:
[rule.dot-imports]
arguments = [{ allowedPackages = ["github.com/onsi/ginkgo/v2","github.com/onsi/gomega"] }]
Description: It is possible to unintentionally import the same package twice. This rule looks for packages that are imported two or more times.
Configuration: N/A
Description: In Go it is idiomatic to minimize nesting statements, a typical example is to avoid if-then-else constructions. This rule spots constructions like
if cond {
// do something
} else {
// do other thing
return ...
}
where the if
condition may be inverted in order to reduce nesting:
if !cond {
// do other thing
return ...
}
// do something
Configuration: ([]string) rule flags. Available flags are:
- preserveScope: do not suggest refactorings that would increase variable scope
Example:
[rule.early-return]
arguments = ["preserveScope"]
Description: Empty blocks make code less readable and could be a symptom of a bug or unfinished refactoring.
Configuration: N/A
Description: Sometimes gofmt
is not enough to enforce a common formatting of a code-base; this rule warns when there are heading or trailing newlines in code blocks.
Configuration: N/A
Description: By convention, for the sake of readability, variables of type error
must be named with the prefix err
.
Configuration: N/A
Description: By convention, for the sake of readability, the errors should be last in the list of returned values by a function.
Configuration: N/A
Description: By convention, for better readability, error messages should not be capitalized or end with punctuation or a newline.
More information here
Configuration: N/A
Description: It is possible to get a simpler program by replacing errors.New(fmt.Sprintf())
with fmt.Errorf()
. This rule spots that kind of simplification opportunities.
Configuration: N/A
Description: This rule enforces consistent usage of make(map[type]type)
or map[type]type{}
for map initialization. It does not affect make(map[type]type, size)
constructions as well as map[type]type{k1: v1}
.
Configuration: (string) Specifies the enforced style for map initialization. The options are:
- "any": No enforcement (default).
- "make": Enforces the usage of
make(map[type]type)
. - "literal": Enforces the usage of
map[type]type{}
.
Example:
[rule.enforce-map-style]
arguments = ["make"]
Description: This rule is designed to maintain consistency in the declaration of repeated argument and return value types in Go functions. It supports three styles: 'any', 'short', and 'full'. The 'any' style is lenient and allows any form of type declaration. The 'short' style encourages omitting repeated types for conciseness, whereas the 'full' style mandates explicitly stating the type for each argument and return value, even if they are repeated, promoting clarity.
Configuration (1): (string) as a single string, it configures both argument and return value styles. Accepts 'any', 'short', or 'full' (default: 'any').
Configuration (2): (map[string]any) as a map, allows separate configuration for function arguments and return values. Valid keys are "funcArgStyle" and "funcRetValStyle", each accepting 'any', 'short', or 'full'. If a key is not specified, the default value of 'any' is used.
Note: The rule applies checks based on the specified styles. For 'full' style, it flags instances where types are omitted in repeated arguments or return values. For 'short' style, it highlights opportunities to omit repeated types for brevity. Incorrect or unknown configuration values will result in an error.
Example (1):
[rule.enforce-repeated-arg-type-style]
arguments = ["short"]
Example (2):
[rule.enforce-repeated-arg-type-style]
arguments = [{ funcArgStyle = "full", funcRetValStyle = "short" }]
Description: This rule enforces consistent usage of make([]type, 0)
, []type{}
, or var []type
for slice initialization.
It does not affect make([]type, non_zero_len, or_non_zero_cap)
constructions as well as []type{v1}
.
Nil slices are always permitted.
Configuration: (string) Specifies the enforced style for slice initialization. The options are:
- "any": No enforcement (default).
- "make": Enforces the usage of
make([]type, 0)
. - "literal": Enforces the usage of
[]type{}
. - "nil": Enforces the usage of
var []type
.
Example:
[rule.enforce-slice-style]
arguments = ["make"]
Description: Exported function and methods should have comments. This warns on undocumented exported functions and methods.
More information here
Configuration: ([]string) rule flags.
Please notice that without configuration, the default behavior of the rule is that of its golint
counterpart.
Available flags are:
- checkPrivateReceivers enables checking public methods of private types
- disableStutteringCheck disables checking for method names that stutter with the package name (i.e. avoid failure messages of the form type name will be used as x.XY by other packages, and that stutters; consider calling this Y)
- sayRepetitiveInsteadOfStutters replaces the use of the term stutters by repetitive in failure messages
Example:
[rule.exported]
arguments = ["checkPrivateReceivers", "disableStutteringCheck"]
Description: This rule helps to enforce a common header for all source files in a project by spotting those files that do not have the specified header.
Configuration: (string) the header to look for in source files.
Example:
[rule.file-header]
arguments = ["This is the text that must appear at the top of source files."]
Description: If a function controls the flow of another by passing it information on what to do, both functions are said to be control-coupled. Coupling among functions must be minimized for better maintainability of the code. This rule warns on boolean parameters that create a control coupling.
Configuration: N/A
Description: Functions returning too many results can be hard to understand/use.
Configuration: (int) the maximum allowed return values
Example:
[rule.function-result-limit]
arguments = [3]
Description: Functions too long (with many statements and/or lines) can be hard to understand.
Configuration: (int,int) the maximum allowed statements and lines. Must be non-negative integers. Set to 0 to disable the check
Example:
[rule.function-length]
arguments = [10, 0]
Will check for functions exceeding 10 statements and will not check the number of lines of functions
Description: Typically, functions with names prefixed with Get are supposed to return a value.
Configuration: N/A
Description: an if-then-else
conditional with identical implementations in both branches is an error.
Configuration: N/A
Description: Checking if an error is nil to just after return the error or nil is redundant.
Configuration: N/A
Description: Aligns with Go's naming conventions, as outlined in the official blog post. It enforces clear and lowercase import alias names, echoing the principles of good package naming. Users can follow these guidelines by default or define a custom regex rule. Importantly, aliases with underscores ("_") are always allowed.
Configuration (1): (string
) as plain string accepts allow regexp pattern for aliases (default: ^[a-z][a-z0-9]{0,}$
).
Configuration (2): (map[string]string
) as a map accepts two values:
- for a key "allowRegex" accepts allow regexp pattern
- for a key "denyRegex deny regexp pattern
Note: If both allowRegex
and denyRegex
are provided, the alias must comply with both of them.
If none are given (i.e. an empty map), the default value ^[a-z][a-z0-9]{0,}$
for allowRegex is used.
Unknown keys will result in an error.
Example (1):
[rule.import-alias-naming]
arguments = ["^[a-z][a-z0-9]{0,}$"]
Example (2):
[rule.import-alias-naming]
arguments = [{ allowRegex = "^[a-z][a-z0-9]{0,}$", denyRegex = '^v\d+$' }]
Description: In Go it is possible to declare identifiers (packages, structs, interfaces, parameters, receivers, variables, constants...) that conflict with the name of an imported package. This rule spots identifiers that shadow an import.
Configuration: N/A
Description: Warns when importing block-listed packages.
Configuration: block-list of package names (or regular expression package names).
Example:
[rule.imports-blocklist]
arguments = ["crypto/md5", "crypto/sha1", "crypto/**/pkix"]
Description: By convention, for better readability, incrementing an integer variable by 1 is recommended to be done using the ++
operator.
This rule spots expressions like i += 1
and i -= 1
and proposes to change them into i++
and i--
.
Configuration: N/A
Description: To improve the readability of code, it is recommended to reduce the indentation as much as possible. This rule highlights redundant else-blocks that can be eliminated from the code.
More information here
Configuration: ([]string) rule flags. Available flags are:
- preserveScope: do not suggest refactorings that would increase variable scope
Example:
[rule.indent-error-flow]
arguments = ["preserveScope"]
Description: Warns in the presence of code lines longer than a configured maximum.
Configuration: (int) maximum line length in characters.
Example:
[rule.line-length-limit]
arguments = [80]
Description: Warns if nesting level of control structures (if-then-else
, for
, switch
) exceeds a given maximum.
Configuration: (int) maximum accepted nesting level of control structures (defaults to 5)
Example:
[max-control-nesting]
arguments = [3]
Description: Packages declaring too many public structs can be hard to understand/use, and could be a symptom of bad design.
This rule warns on files declaring more than a configured, maximum number of public structs.
Configuration: (int) the maximum allowed public structs
Example:
[rule.max-public-structs]
arguments = [3]
Description: A function that modifies its parameters can be hard to understand. It can also be misleading if the arguments are passed by value by the caller. This rule warns when a function modifies one or more of its parameters.
Configuration: N/A
Description: A method that modifies its receiver value can have undesired behavior. The modification can be also the root of a bug because the actual value receiver could be a copy of that used at the calling site. This rule warns when a method modifies its receiver.
Configuration: N/A
Description: Packages declaring structs that contain other inline struct definitions can be hard to understand/read for other developers.
Configuration: N/A
Description: conditional expressions can be written to take advantage of short circuit evaluation and speed up its average evaluation time by forcing the evaluation of less time-consuming terms before more costly ones. This rule spots logical expressions where the order of evaluation of terms seems non optimal. Please notice that confidence of this rule is low and is up to the user to decide if the suggested rewrite of the expression keeps the semantics of the original one.
Configuration: N/A
Example:
if isGenerated(content) && !config.IgnoreGeneratedHeader {
Swap left and right side :
if !config.IgnoreGeneratedHeader && isGenerated(content) {
Description: Packages should have comments. This rule warns on undocumented packages and when packages comments are detached to the package
keyword.
More information here
Configuration: N/A
Description: Range variables in a loop are reused at each iteration. This rule warns when assigning the address of the variable, passing the address to append() or using it in a map.
Configuration: N/A
Description: Range variables in a loop are reused at each iteration; therefore a goroutine created in a loop will point to the range variable with from the upper scope. This way, the goroutine could use the variable with an undesired value. This rule warns when a range value (or index) is used inside a closure
Configuration: N/A
Description: This rule suggests a shorter way of writing ranges that do not use the second value.
Configuration: N/A
Description: By convention, receiver names in a method should reflect their identity. For example, if the receiver is of type Parts
, p
is an adequate name for it. Contrary to other languages, it is not idiomatic to name receivers as this
or self
.
Configuration: N/A
Description: Constant names like false
, true
, nil
, function names like append
, make
, and basic type names like bool
, and byte
are not reserved words of the language; therefore the can be redefined.
Even if possible, redefining these built in names can lead to bugs very difficult to detect.
Configuration: N/A
Description: This rule warns on redundant import aliases. This happens when the alias used on the import statement matches the imported package name.
Configuration: N/A
Description: This rule allows you to configure a list of regular expressions that string literals in certain function calls are checked against. This is geared towards user facing applications where string literals are often used for messages that will be presented to users, so it may be desirable to enforce consistent formatting.
Configuration: Each argument is a slice containing 2-3 strings: a scope, a regex, and an optional error message.
-
The first string defines a scope. This controls which string literals the regex will apply to, and is defined as a function argument. It must contain at least a function name (
core.WriteError
). Scopes may optionally contain a number specifying which argument in the function to check (core.WriteError[1]
), as well as a struct field (core.WriteError[1].Message
, only works for top level fields). Function arguments are counted starting at 0, so[0]
would refer to the first argument,[1]
would refer to the second, etc. If no argument number is provided, the first argument will be used (same as[0]
). -
The second string is a regular expression (beginning and ending with a
/
character), which will be used to check the string literals in the scope. The default semantics is "strings matching the regular expression are OK". If you need to inverse the semantics you can add a!
just before the first/
. Examples:- with
"/^[A-Z].*$/"
the rule will accept strings starting with capital letters - with
"!/^[A-Z].*$/"
the rule will a fail on strings starting with capital letters
- with
-
The third string (optional) is a message containing the purpose for the regex, which will be used in lint errors.
Example:
[rule.string-format]
arguments = [
["core.WriteError[1].Message", "/^([^A-Z]|$)/", "must not start with a capital letter"],
["fmt.Errorf[0]", "/(^|[^\\.!?])$/", "must not end in punctuation"],
["panic", "/^[^\\n]*$/", "must not contain line breaks"],
]
Description: explicit type conversion string(i)
where i
has an integer type other than rune
might behave not as expected by the developer (e.g. string(42)
is not "42"
). This rule spot that kind of suspicious conversions.
Configuration: N/A
Description: Struct tags are not checked at compile time. This rule, checks and warns if it finds errors in common struct tags types like: asn1, default, json, protobuf, xml, yaml.
Configuration: (optional) list of user defined options.
Example:
To accept the inline
option in JSON tags (and outline
and gnu
in BSON tags) you must provide the following configuration
[rule.struct-tag]
arguments = ["json,inline", "bson,outline,gnu"]
Description: To improve the readability of code, it is recommended to reduce the indentation as much as possible. This rule highlights redundant else-blocks that can be eliminated from the code.
Configuration: ([]string) rule flags. Available flags are:
- preserveScope: do not suggest refactorings that would increase variable scope
Example:
[rule.superfluous-else]
arguments = ["preserveScope"]
Description: This rule warns when using ==
and !=
for equality check time.Time
and suggest to time.time.Equal
method, for about information follow this link
Configuration: N/A
Description: Using unit-specific suffix like "Secs", "Mins", ... when naming variables of type time.Duration
can be misleading, this rule highlights those cases.
Configuration: N/A
Description: This rule checks whether a type assertion result is checked (the ok
value), preventing unexpected panic
s.
Configuration: list of key-value-pair-map ([]map[string]any
).
acceptIgnoredAssertionResult
: (bool) defaultfalse
, set it totrue
to accept ignored type assertion results like this:
foo, _ := bar(.*Baz).
// ^
Example:
[rule.unchecked-type-assertion]
arguments = [{acceptIgnoredAssertionResult=true}]
Description: Unconditional recursive calls will produce infinite recursion, thus program stack overflow. This rule detects and warns about unconditional (direct) recursive calls.
Configuration: N/A
Description: this rule warns on wrongly named un-exported symbols, i.e. un-exported symbols whose name start with a capital letter.
Configuration: N/A
Description: This rule warns when an exported function or method returns a value of an un-exported type.
Configuration: N/A
Description: This rule warns when errors returned by a function are not explicitly handled on the caller side.
Configuration: function names regexp patterns to ignore
Example:
[unhandled-error]
arguments = ["os\.(Create|WriteFile|Chmod)", "fmt\.Print", "myFunction", "net\..*", "bytes\.Buffer\.Write"]
Description: This rule suggests to remove redundant statements like a break
at the end of a case block, for improving the code's readability.
Configuration: N/A
Description: This rule spots and proposes to remove unreachable code.
Configuration: N/A
Description: This rule warns on unused parameters. Functions or methods with unused parameters can be a symptom of an unfinished refactoring or a bug.
Configuration: Supports arguments with single of map[string]any
with option allowRegex
to provide additional to _
mask to allowed unused parameter names, for example:
[rule.unused-parameter]
Arguments = [{ allowRegex = "^_" }]
allows any names started with _
, not just _
itself:
func SomeFunc(_someObj *MyStruct) {} // matches rule
Description: This rule warns on unused method receivers. Methods with unused receivers can be a symptom of an unfinished refactoring or a bug.
Configuration: Supports arguments with single of map[string]any
with option allowRegex
to provide additional to _
mask to allowed unused receiver names, for example:
[rule.unused-receiver]
Arguments = [{ allowRegex = "^_" }]
allows any names started with _
, not just _
itself:
func (_my *MyStruct) SomeMethod() {} // matches rule
Description: Since Go 1.18, interface{}
has an alias: any
. This rule proposes to replace instances of interface{}
with any
.
Configuration: N/A
Description: This rule warns on useless break
statements in case clauses of switch and select statements. Go, unlike other programming languages like C, only executes statements of the selected case while ignoring the subsequent case clauses.
Therefore, inserting a break
at the end of a case clause has no effect.
Because break
statements are rarely used in case clauses, when switch or select statements are inside a for-loop, the programmer might wrongly assume that a break
in a case clause will take the control out of the loop.
The rule emits a specific warning for such cases.
Configuration: N/A
Description: This rule proposes simplifications of variable declarations.
Configuration: N/A
Description: This rule warns when initialism, variable or package naming conventions are not followed.
Configuration: This rule accepts two slices of strings and one optional slice with single map with named parameters.
(it's due to TOML hasn't "slice of any" and we keep backward compatibility with previous config version)
First slice is an allowlist and second one is a blocklist of initialisms.
In map, you can add "upperCaseConst=true" parameter to allow UPPER_CASE
for const
By default, the rule behaves exactly as the alternative in golint
but optionally, you can relax it (see golint/lint/issues/89)
Example:
[rule.var-naming]
arguments = [["ID"], ["VM"], [{upperCaseConst=true}]]
You can also add "skipPackageNameChecks=true" to skip package name checks.
Example:
[rule.var-naming]
arguments = [[], [], [{skipPackageNameChecks=true}]]
Description: Function parameters that are passed by value, are in fact a copy of the original argument. Passing a copy of a sync.WaitGroup
is usually not what the developer wants to do.
This rule warns when a sync.WaitGroup
expected as a by-value parameter in a function or method.
Configuration: N/A