VictoriaMetrics/vendor/honnef.co/go/tools/staticcheck/doc.go
Aliaksandr Valialkin b459919250 make vendor-update
2020-02-26 20:45:27 +02:00

881 lines
25 KiB
Go
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

package staticcheck
import "honnef.co/go/tools/lint"
var Docs = map[string]*lint.Documentation{
"SA1000": {
Title: `Invalid regular expression`,
Since: "2017.1",
},
"SA1001": {
Title: `Invalid template`,
Since: "2017.1",
},
"SA1002": {
Title: `Invalid format in time.Parse`,
Since: "2017.1",
},
"SA1003": {
Title: `Unsupported argument to functions in encoding/binary`,
Text: `The encoding/binary package can only serialize types with known sizes.
This precludes the use of the int and uint types, as their sizes
differ on different architectures. Furthermore, it doesn't support
serializing maps, channels, strings, or functions.
Before Go 1.8, bool wasn't supported, either.`,
Since: "2017.1",
},
"SA1004": {
Title: `Suspiciously small untyped constant in time.Sleep`,
Text: `The time.Sleep function takes a time.Duration as its only argument.
Durations are expressed in nanoseconds. Thus, calling time.Sleep(1)
will sleep for 1 nanosecond. This is a common source of bugs, as sleep
functions in other languages often accept seconds or milliseconds.
The time package provides constants such as time.Second to express
large durations. These can be combined with arithmetic to express
arbitrary durations, for example '5 * time.Second' for 5 seconds.
If you truly meant to sleep for a tiny amount of time, use
'n * time.Nanosecond' to signal to Staticcheck that you did mean to sleep
for some amount of nanoseconds.`,
Since: "2017.1",
},
"SA1005": {
Title: `Invalid first argument to exec.Command`,
Text: `os/exec runs programs directly (using variants of the fork and exec
system calls on Unix systems). This shouldn't be confused with running
a command in a shell. The shell will allow for features such as input
redirection, pipes, and general scripting. The shell is also
responsible for splitting the user's input into a program name and its
arguments. For example, the equivalent to
ls / /tmp
would be
exec.Command("ls", "/", "/tmp")
If you want to run a command in a shell, consider using something like
the following but be aware that not all systems, particularly
Windows, will have a /bin/sh program:
exec.Command("/bin/sh", "-c", "ls | grep Awesome")`,
Since: "2017.1",
},
"SA1006": {
Title: `Printf with dynamic first argument and no further arguments`,
Text: `Using fmt.Printf with a dynamic first argument can lead to unexpected
output. The first argument is a format string, where certain character
combinations have special meaning. If, for example, a user were to
enter a string such as
Interest rate: 5%
and you printed it with
fmt.Printf(s)
it would lead to the following output:
Interest rate: 5%!(NOVERB).
Similarly, forming the first parameter via string concatenation with
user input should be avoided for the same reason. When printing user
input, either use a variant of fmt.Print, or use the %s Printf verb
and pass the string as an argument.`,
Since: "2017.1",
},
"SA1007": {
Title: `Invalid URL in net/url.Parse`,
Since: "2017.1",
},
"SA1008": {
Title: `Non-canonical key in http.Header map`,
Text: `Keys in http.Header maps are canonical, meaning they follow a specific
combination of uppercase and lowercase letters. Methods such as
http.Header.Add and http.Header.Del convert inputs into this canonical
form before manipulating the map.
When manipulating http.Header maps directly, as opposed to using the
provided methods, care should be taken to stick to canonical form in
order to avoid inconsistencies. The following piece of code
demonstrates one such inconsistency:
h := http.Header{}
h["etag"] = []string{"1234"}
h.Add("etag", "5678")
fmt.Println(h)
// Output:
// map[Etag:[5678] etag:[1234]]
The easiest way of obtaining the canonical form of a key is to use
http.CanonicalHeaderKey.`,
Since: "2017.1",
},
"SA1010": {
Title: `(*regexp.Regexp).FindAll called with n == 0, which will always return zero results`,
Text: `If n >= 0, the function returns at most n matches/submatches. To
return all results, specify a negative number.`,
Since: "2017.1",
},
"SA1011": {
Title: `Various methods in the strings package expect valid UTF-8, but invalid input is provided`,
Since: "2017.1",
},
"SA1012": {
Title: `A nil context.Context is being passed to a function, consider using context.TODO instead`,
Since: "2017.1",
},
"SA1013": {
Title: `io.Seeker.Seek is being called with the whence constant as the first argument, but it should be the second`,
Since: "2017.1",
},
"SA1014": {
Title: `Non-pointer value passed to Unmarshal or Decode`,
Since: "2017.1",
},
"SA1015": {
Title: `Using time.Tick in a way that will leak. Consider using time.NewTicker, and only use time.Tick in tests, commands and endless functions`,
Since: "2017.1",
},
"SA1016": {
Title: `Trapping a signal that cannot be trapped`,
Text: `Not all signals can be intercepted by a process. Speficially, on
UNIX-like systems, the syscall.SIGKILL and syscall.SIGSTOP signals are
never passed to the process, but instead handled directly by the
kernel. It is therefore pointless to try and handle these signals.`,
Since: "2017.1",
},
"SA1017": {
Title: `Channels used with os/signal.Notify should be buffered`,
Text: `The os/signal package uses non-blocking channel sends when delivering
signals. If the receiving end of the channel isn't ready and the
channel is either unbuffered or full, the signal will be dropped. To
avoid missing signals, the channel should be buffered and of the
appropriate size. For a channel used for notification of just one
signal value, a buffer of size 1 is sufficient.`,
Since: "2017.1",
},
"SA1018": {
Title: `strings.Replace called with n == 0, which does nothing`,
Text: `With n == 0, zero instances will be replaced. To replace all
instances, use a negative number, or use strings.ReplaceAll.`,
Since: "2017.1",
},
"SA1019": {
Title: `Using a deprecated function, variable, constant or field`,
Since: "2017.1",
},
"SA1020": {
Title: `Using an invalid host:port pair with a net.Listen-related function`,
Since: "2017.1",
},
"SA1021": {
Title: `Using bytes.Equal to compare two net.IP`,
Text: `A net.IP stores an IPv4 or IPv6 address as a slice of bytes. The
length of the slice for an IPv4 address, however, can be either 4 or
16 bytes long, using different ways of representing IPv4 addresses. In
order to correctly compare two net.IPs, the net.IP.Equal method should
be used, as it takes both representations into account.`,
Since: "2017.1",
},
"SA1023": {
Title: `Modifying the buffer in an io.Writer implementation`,
Text: `Write must not modify the slice data, even temporarily.`,
Since: "2017.1",
},
"SA1024": {
Title: `A string cutset contains duplicate characters`,
Text: `The strings.TrimLeft and strings.TrimRight functions take cutsets, not
prefixes. A cutset is treated as a set of characters to remove from a
string. For example,
strings.TrimLeft("42133word", "1234"))
will result in the string "word" any characters that are 1, 2, 3 or
4 are cut from the left of the string.
In order to remove one string from another, use strings.TrimPrefix instead.`,
Since: "2017.1",
},
"SA1025": {
Title: `It is not possible to use (*time.Timer).Reset's return value correctly`,
Since: "2019.1",
},
"SA1026": {
Title: `Cannot marshal channels or functions`,
Since: "2019.2",
},
"SA1027": {
Title: `Atomic access to 64-bit variable must be 64-bit aligned`,
Text: `On ARM, x86-32, and 32-bit MIPS, it is the caller's responsibility to
arrange for 64-bit alignment of 64-bit words accessed atomically. The
first word in a variable or in an allocated struct, array, or slice
can be relied upon to be 64-bit aligned.
You can use the structlayout tool to inspect the alignment of fields
in a struct.`,
Since: "2019.2",
},
"SA1028": {
Title: `sort.Slice can only be used on slices`,
Text: `The first argument of sort.Slice must be a slice.`,
Since: "2020.1",
},
"SA1029": {
Title: `Inappropriate key in call to context.WithValue`,
Text: `The provided key must be comparable and should not be
of type string or any other built-in type to avoid collisions between
packages using context. Users of WithValue should define their own
types for keys.
To avoid allocating when assigning to an interface{},
context keys often have concrete type struct{}. Alternatively,
exported context key variables' static type should be a pointer or
interface.`,
Since: "2020.1",
},
"SA2000": {
Title: `sync.WaitGroup.Add called inside the goroutine, leading to a race condition`,
Since: "2017.1",
},
"SA2001": {
Title: `Empty critical section, did you mean to defer the unlock?`,
Text: `Empty critical sections of the kind
mu.Lock()
mu.Unlock()
are very often a typo, and the following was intended instead:
mu.Lock()
defer mu.Unlock()
Do note that sometimes empty critical sections can be useful, as a
form of signaling to wait on another goroutine. Many times, there are
simpler ways of achieving the same effect. When that isn't the case,
the code should be amply commented to avoid confusion. Combining such
comments with a //lint:ignore directive can be used to suppress this
rare false positive.`,
Since: "2017.1",
},
"SA2002": {
Title: `Called testing.T.FailNow or SkipNow in a goroutine, which isn't allowed`,
Since: "2017.1",
},
"SA2003": {
Title: `Deferred Lock right after locking, likely meant to defer Unlock instead`,
Since: "2017.1",
},
"SA3000": {
Title: `TestMain doesn't call os.Exit, hiding test failures`,
Text: `Test executables (and in turn 'go test') exit with a non-zero status
code if any tests failed. When specifying your own TestMain function,
it is your responsibility to arrange for this, by calling os.Exit with
the correct code. The correct code is returned by (*testing.M).Run, so
the usual way of implementing TestMain is to end it with
os.Exit(m.Run()).`,
Since: "2017.1",
},
"SA3001": {
Title: `Assigning to b.N in benchmarks distorts the results`,
Text: `The testing package dynamically sets b.N to improve the reliability of
benchmarks and uses it in computations to determine the duration of a
single operation. Benchmark code must not alter b.N as this would
falsify results.`,
Since: "2017.1",
},
"SA4000": {
Title: `Boolean expression has identical expressions on both sides`,
Since: "2017.1",
},
"SA4001": {
Title: `&*x gets simplified to x, it does not copy x`,
Since: "2017.1",
},
"SA4002": {
Title: `Comparing strings with known different sizes has predictable results`,
Since: "2017.1",
},
"SA4003": {
Title: `Comparing unsigned values against negative values is pointless`,
Since: "2017.1",
},
"SA4004": {
Title: `The loop exits unconditionally after one iteration`,
Since: "2017.1",
},
"SA4005": {
Title: `Field assignment that will never be observed. Did you mean to use a pointer receiver?`,
Since: "2017.1",
},
"SA4006": {
Title: `A value assigned to a variable is never read before being overwritten. Forgotten error check or dead code?`,
Since: "2017.1",
},
"SA4008": {
Title: `The variable in the loop condition never changes, are you incrementing the wrong variable?`,
Since: "2017.1",
},
"SA4009": {
Title: `A function argument is overwritten before its first use`,
Since: "2017.1",
},
"SA4010": {
Title: `The result of append will never be observed anywhere`,
Since: "2017.1",
},
"SA4011": {
Title: `Break statement with no effect. Did you mean to break out of an outer loop?`,
Since: "2017.1",
},
"SA4012": {
Title: `Comparing a value against NaN even though no value is equal to NaN`,
Since: "2017.1",
},
"SA4013": {
Title: `Negating a boolean twice (!!b) is the same as writing b. This is either redundant, or a typo.`,
Since: "2017.1",
},
"SA4014": {
Title: `An if/else if chain has repeated conditions and no side-effects; if the condition didn't match the first time, it won't match the second time, either`,
Since: "2017.1",
},
"SA4015": {
Title: `Calling functions like math.Ceil on floats converted from integers doesn't do anything useful`,
Since: "2017.1",
},
"SA4016": {
Title: `Certain bitwise operations, such as x ^ 0, do not do anything useful`,
Since: "2017.1",
},
"SA4017": {
Title: `A pure function's return value is discarded, making the call pointless`,
Since: "2017.1",
},
"SA4018": {
Title: `Self-assignment of variables`,
Since: "2017.1",
},
"SA4019": {
Title: `Multiple, identical build constraints in the same file`,
Since: "2017.1",
},
"SA4020": {
Title: `Unreachable case clause in a type switch`,
Text: `In a type switch like the following
type T struct{}
func (T) Read(b []byte) (int, error) { return 0, nil }
var v interface{} = T{}
switch v.(type) {
case io.Reader:
// ...
case T:
// unreachable
}
the second case clause can never be reached because T implements
io.Reader and case clauses are evaluated in source order.
Another example:
type T struct{}
func (T) Read(b []byte) (int, error) { return 0, nil }
func (T) Close() error { return nil }
var v interface{} = T{}
switch v.(type) {
case io.Reader:
// ...
case io.ReadCloser:
// unreachable
}
Even though T has a Close method and thus implements io.ReadCloser,
io.Reader will always match first. The method set of io.Reader is a
subset of io.ReadCloser. Thus it is impossible to match the second
case without matching the first case.
Structurally equivalent interfaces
A special case of the previous example are structurally identical
interfaces. Given these declarations
type T error
type V error
func doSomething() error {
err, ok := doAnotherThing()
if ok {
return T(err)
}
return U(err)
}
the following type switch will have an unreachable case clause:
switch doSomething().(type) {
case T:
// ...
case V:
// unreachable
}
T will always match before V because they are structurally equivalent
and therefore doSomething()'s return value implements both.`,
Since: "2019.2",
},
"SA4021": {
Title: `x = append(y) is equivalent to x = y`,
Since: "2019.2",
},
"SA4022": {
Title: `Comparing the address of a variable against nil`,
Text: `Code such as 'if &x == nil' is meaningless, because taking the address of a variable always yields a non-nil pointer.`,
Since: "2020.1",
},
"SA5000": {
Title: `Assignment to nil map`,
Since: "2017.1",
},
"SA5001": {
Title: `Defering Close before checking for a possible error`,
Since: "2017.1",
},
"SA5002": {
Title: `The empty for loop (for {}) spins and can block the scheduler`,
Since: "2017.1",
},
"SA5003": {
Title: `Defers in infinite loops will never execute`,
Text: `Defers are scoped to the surrounding function, not the surrounding
block. In a function that never returns, i.e. one containing an
infinite loop, defers will never execute.`,
Since: "2017.1",
},
"SA5004": {
Title: `for { select { ... with an empty default branch spins`,
Since: "2017.1",
},
"SA5005": {
Title: `The finalizer references the finalized object, preventing garbage collection`,
Text: `A finalizer is a function associated with an object that runs when the
garbage collector is ready to collect said object, that is when the
object is no longer referenced by anything.
If the finalizer references the object, however, it will always remain
as the final reference to that object, preventing the garbage
collector from collecting the object. The finalizer will never run,
and the object will never be collected, leading to a memory leak. That
is why the finalizer should instead use its first argument to operate
on the object. That way, the number of references can temporarily go
to zero before the object is being passed to the finalizer.`,
Since: "2017.1",
},
"SA5006": {
Title: `Slice index out of bounds`,
Since: "2017.1",
},
"SA5007": {
Title: `Infinite recursive call`,
Text: `A function that calls itself recursively needs to have an exit
condition. Otherwise it will recurse forever, until the system runs
out of memory.
This issue can be caused by simple bugs such as forgetting to add an
exit condition. It can also happen "on purpose". Some languages have
tail call optimization which makes certain infinite recursive calls
safe to use. Go, however, does not implement TCO, and as such a loop
should be used instead.`,
Since: "2017.1",
},
"SA5008": {
Title: `Invalid struct tag`,
Since: "2019.2",
},
"SA5009": {
Title: `Invalid Printf call`,
Since: "2019.2",
},
"SA5010": {
Title: `Impossible type assertion`,
Text: `Some type assertions can be statically proven to be
impossible. This is the case when the method sets of both
arguments of the type assertion conflict with each other, for
example by containing the same method with different
signatures.
The Go compiler already applies this check when asserting from an
interface value to a concrete type. If the concrete type misses
methods from the interface, or if function signatures don't match,
then the type assertion can never succeed.
This check applies the same logic when asserting from one interface to
another. If both interface types contain the same method but with
different signatures, then the type assertion can never succeed,
either.`,
Since: "2020.1",
},
"SA5011": {
Title: `Possible nil pointer dereference`,
Text: `A pointer is being dereferenced unconditionally, while
also being checked against nil in another place. This suggests that
the pointer may be nil and dereferencing it may panic. This is
commonly a result of improperly ordered code or missing return
statements. Consider the following examples:
func fn(x *int) {
fmt.Println(*x)
// This nil check is equally important for the previous dereference
if x != nil {
foo(*x)
}
}
func TestFoo(t *testing.T) {
x := compute()
if x == nil {
t.Errorf("nil pointer received")
}
// t.Errorf does not abort the test, so if x is nil, the next line will panic.
foo(*x)
}
Staticcheck tries to deduce which functions abort control flow.
For example, it is aware that a function will not continue
execution after a call to panic or log.Fatal. However, sometimes
this detection fails, in particular in the presence of
conditionals. Consider the following example:
func Log(msg string, level int) {
fmt.Println(msg)
if level == levelFatal {
os.Exit(1)
}
}
func Fatal(msg string) {
Log(msg, levelFatal)
}
func fn(x *int) {
if x == nil {
Fatal("unexpected nil pointer")
}
fmt.Println(*x)
}
Staticcheck will flag the dereference of x, even though it is perfectly
safe. Staticcheck is not able to deduce that a call to
Fatal will exit the program. For the time being, the easiest
workaround is to modify the definition of Fatal like so:
func Fatal(msg string) {
Log(msg, levelFatal)
panic("unreachable")
}
We also hard-code functions from common logging packages such as
logrus. Please file an issue if we're missing support for a
popular package.`,
Since: "2020.1",
},
"SA6000": {
Title: `Using regexp.Match or related in a loop, should use regexp.Compile`,
Since: "2017.1",
},
"SA6001": {
Title: `Missing an optimization opportunity when indexing maps by byte slices`,
Text: `Map keys must be comparable, which precludes the use of byte slices.
This usually leads to using string keys and converting byte slices to
strings.
Normally, a conversion of a byte slice to a string needs to copy the data and
causes allocations. The compiler, however, recognizes m[string(b)] and
uses the data of b directly, without copying it, because it knows that
the data can't change during the map lookup. This leads to the
counter-intuitive situation that
k := string(b)
println(m[k])
println(m[k])
will be less efficient than
println(m[string(b)])
println(m[string(b)])
because the first version needs to copy and allocate, while the second
one does not.
For some history on this optimization, check out commit
f5f5a8b6209f84961687d993b93ea0d397f5d5bf in the Go repository.`,
Since: "2017.1",
},
"SA6002": {
Title: `Storing non-pointer values in sync.Pool allocates memory`,
Text: `A sync.Pool is used to avoid unnecessary allocations and reduce the
amount of work the garbage collector has to do.
When passing a value that is not a pointer to a function that accepts
an interface, the value needs to be placed on the heap, which means an
additional allocation. Slices are a common thing to put in sync.Pools,
and they're structs with 3 fields (length, capacity, and a pointer to
an array). In order to avoid the extra allocation, one should store a
pointer to the slice instead.
See the comments on https://go-review.googlesource.com/c/go/+/24371
that discuss this problem.`,
Since: "2017.1",
},
"SA6003": {
Title: `Converting a string to a slice of runes before ranging over it`,
Text: `You may want to loop over the runes in a string. Instead of converting
the string to a slice of runes and looping over that, you can loop
over the string itself. That is,
for _, r := range s {}
and
for _, r := range []rune(s) {}
will yield the same values. The first version, however, will be faster
and avoid unnecessary memory allocations.
Do note that if you are interested in the indices, ranging over a
string and over a slice of runes will yield different indices. The
first one yields byte offsets, while the second one yields indices in
the slice of runes.`,
Since: "2017.1",
},
"SA6005": {
Title: `Inefficient string comparison with strings.ToLower or strings.ToUpper`,
Text: `Converting two strings to the same case and comparing them like so
if strings.ToLower(s1) == strings.ToLower(s2) {
...
}
is significantly more expensive than comparing them with
strings.EqualFold(s1, s2). This is due to memory usage as well as
computational complexity.
strings.ToLower will have to allocate memory for the new strings, as
well as convert both strings fully, even if they differ on the very
first byte. strings.EqualFold, on the other hand, compares the strings
one character at a time. It doesn't need to create two intermediate
strings and can return as soon as the first non-matching character has
been found.
For a more in-depth explanation of this issue, see
https://blog.digitalocean.com/how-to-efficiently-compare-strings-in-go/`,
Since: "2019.2",
},
"SA9001": {
Title: `Defers in range loops may not run when you expect them to`,
Since: "2017.1",
},
"SA9002": {
Title: `Using a non-octal os.FileMode that looks like it was meant to be in octal.`,
Since: "2017.1",
},
"SA9003": {
Title: `Empty body in an if or else branch`,
Since: "2017.1",
},
"SA9004": {
Title: `Only the first constant has an explicit type`,
Text: `In a constant declaration such as the following:
const (
First byte = 1
Second = 2
)
the constant Second does not have the same type as the constant First.
This construct shouldn't be confused with
const (
First byte = iota
Second
)
where First and Second do indeed have the same type. The type is only
passed on when no explicit value is assigned to the constant.
When declaring enumerations with explicit values it is therefore
important not to write
const (
EnumFirst EnumType = 1
EnumSecond = 2
EnumThird = 3
)
This discrepancy in types can cause various confusing behaviors and
bugs.
Wrong type in variable declarations
The most obvious issue with such incorrect enumerations expresses
itself as a compile error:
package pkg
const (
EnumFirst uint8 = 1
EnumSecond = 2
)
func fn(useFirst bool) {
x := EnumSecond
if useFirst {
x = EnumFirst
}
}
fails to compile with
./const.go:11:5: cannot use EnumFirst (type uint8) as type int in assignment
Losing method sets
A more subtle issue occurs with types that have methods and optional
interfaces. Consider the following:
package main
import "fmt"
type Enum int
func (e Enum) String() string {
return "an enum"
}
const (
EnumFirst Enum = 1
EnumSecond = 2
)
func main() {
fmt.Println(EnumFirst)
fmt.Println(EnumSecond)
}
This code will output
an enum
2
as EnumSecond has no explicit type, and thus defaults to int.`,
Since: "2019.1",
},
"SA9005": {
Title: `Trying to marshal a struct with no public fields nor custom marshaling`,
Text: `The encoding/json and encoding/xml packages only operate on exported
fields in structs, not unexported ones. It is usually an error to try
to (un)marshal structs that only consist of unexported fields.
This check will not flag calls involving types that define custom
marshaling behavior, e.g. via MarshalJSON methods. It will also not
flag empty structs.`,
Since: "2019.2",
},
}