blob: b1c15295d62e2e4606a55a6f46c09b35b688dc2f [file] [log] [blame]
<!--{
"Title": "Frequently Asked Questions (FAQ)",
"Path": "/doc/faq"
}-->
<h2 id="Origins">Origins</h2>
<h3 id="What_is_the_purpose_of_the_project">
What is the purpose of the project?</h3>
<p>
At the time of Go's inception, only a decade ago, the programming world was different from today.
Production software was usually written in C++ or Java,
GitHub did not exist, most computers were not yet multiprocessors,
and other than Visual Studio and Eclipse there were few IDEs or other high-level tools available
at all, let alone for free on the Internet.
</p>
<p>
Meanwhile, we had become frustrated by the undue complexity required to use
the languages we worked with to develop server software.
Computers had become enormously quicker since languages such as
C, C++ and Java were first developed but the act of programming had not
itself advanced nearly as much.
Also, it was clear that multiprocessors were becoming universal but
most languages offered little help to program them efficiently
and safely.
</p>
<p>
We decided to take a step back and think about what major issues were
going to dominate software engineering in the years ahead as technology
developed, and how a new language might help address them.
For instance, the rise of multicore CPUs argued that a language should
provide first-class support for some sort of concurrency or parallelism.
And to make resource management tractable in a large concurrent program,
garbage collection, or at least some sort of safe automatic memory management was required.
</p>
<p>
These considerations led to
<a href="https://commandcenter.blogspot.com/2017/09/go-ten-years-and-climbing.html">a
series of discussions</a> from which Go arose, first as a set of ideas and
desiderata, then as a language.
An overarching goal was that Go do more to help the working programmer
by enabling tooling, automating mundane tasks such as code formatting,
and removing obstacles to working on large code bases.
</p>
<p>
A much more expansive description of the goals of Go and how
they are met, or at least approached, is available in the article,
<a href="//talks.golang.org/2012/splash.article">Go at Google:
Language Design in the Service of Software Engineering</a>.
</p>
<h3 id="history">
What is the history of the project?</h3>
<p>
Robert Griesemer, Rob Pike and Ken Thompson started sketching the
goals for a new language on the white board on September 21, 2007.
Within a few days the goals had settled into a plan to do something
and a fair idea of what it would be. Design continued part-time in
parallel with unrelated work. By January 2008, Ken had started work
on a compiler with which to explore ideas; it generated C code as its
output. By mid-year the language had become a full-time project and
had settled enough to attempt a production compiler. In May 2008,
Ian Taylor independently started on a GCC front end for Go using the
draft specification. Russ Cox joined in late 2008 and helped move the language
and libraries from prototype to reality.
</p>
<p>
Go became a public open source project on November 10, 2009.
Countless people from the community have contributed ideas, discussions, and code.
</p>
<p>
There are now millions of Go programmers—gophers—around the world,
and there are more every day.
Go's success has far exceeded our expectations.
</p>
<h3 id="gopher">
What's the origin of the gopher mascot?</h3>
<p>
The mascot and logo were designed by
<a href="https://reneefrench.blogspot.com">Renée French</a>, who also designed
<a href="https://9p.io/plan9/glenda.html">Glenda</a>,
the Plan 9 bunny.
A <a href="https://blog.golang.org/gopher">blog post</a>
about the gopher explains how it was
derived from one she used for a <a href="https://wfmu.org/">WFMU</a>
T-shirt design some years ago.
The logo and mascot are covered by the
<a href="https://creativecommons.org/licenses/by/3.0/">Creative Commons Attribution 3.0</a>
license.
</p>
<p>
The gopher has a
<a href="/doc/gopher/modelsheet.jpg">model sheet</a>
illustrating his characteristics and how to represent them correctly.
The model sheet was first shown in a
<a href="https://www.youtube.com/watch?v=4rw_B4yY69k">talk</a>
by Renée at Gophercon in 2016.
He has unique features; he's the <em>Go gopher</em>, not just any old gopher.
</p>
<h3 id="creating_a_new_language">
Why did you create a new language?</h3>
<p>
Go was born out of frustration with existing languages and
environments for the work we were doing at Google.
Programming had become too
difficult and the choice of languages was partly to blame. One had to
choose either efficient compilation, efficient execution, or ease of
programming; all three were not available in the same mainstream
language. Programmers who could were choosing ease over
safety and efficiency by moving to dynamically typed languages such as
Python and JavaScript rather than C++ or, to a lesser extent, Java.
</p>
<p>
We were not alone in our concerns.
After many years with a pretty quiet landscape for programming languages,
Go was among the first of several new languages—Rust,
Elixir, Swift, and more—that have made programming language development
an active, almost mainstream field again.
</p>
<p>
Go addressed these issues by attempting to combine the ease of programming of an interpreted,
dynamically typed
language with the efficiency and safety of a statically typed, compiled language.
It also aimed to be modern, with support for networked and multicore
computing. Finally, working with Go is intended to be <i>fast</i>: it should take
at most a few seconds to build a large executable on a single computer.
To meet these goals required addressing a number of
linguistic issues: an expressive but lightweight type system;
concurrency and garbage collection; rigid dependency specification;
and so on. These cannot be addressed well by libraries or tools; a new
language was called for.
</p>
<p>
The article <a href="//talks.golang.org/2012/splash.article">Go at Google</a>
discusses the background and motivation behind the design of the Go language,
as well as providing more detail about many of the answers presented in this FAQ.
</p>
<h3 id="ancestors">
What are Go's ancestors?</h3>
<p>
Go is mostly in the C family (basic syntax),
with significant input from the Pascal/Modula/Oberon
family (declarations, packages),
plus some ideas from languages
inspired by Tony Hoare's CSP,
such as Newsqueak and Limbo (concurrency).
However, it is a new language across the board.
In every respect the language was designed by thinking
about what programmers do and how to make programming, at least the
kind of programming we do, more effective, which means more fun.
</p>
<h3 id="principles">
What are the guiding principles in the design?</h3>
<p>
When Go was designed, Java and C++ were the most commonly
used languages for writing servers, at least at Google.
We felt that these languages required
too much bookkeeping and repetition.
Some programmers reacted by moving towards more dynamic,
fluid languages like Python, at the cost of efficiency and
type safety.
We felt it should be possible to have the efficiency,
the safety, and the fluidity in a single language.
</p>
<p>
Go attempts to reduce the amount of typing in both senses of the word.
Throughout its design, we have tried to reduce clutter and
complexity. There are no forward declarations and no header files;
everything is declared exactly once. Initialization is expressive,
automatic, and easy to use. Syntax is clean and light on keywords.
Stuttering (<code>foo.Foo* myFoo = new(foo.Foo)</code>) is reduced by
simple type derivation using the <code>:=</code>
declare-and-initialize construct. And perhaps most radically, there
is no type hierarchy: types just <i>are</i>, they don't have to
announce their relationships. These simplifications allow Go to be
expressive yet comprehensible without sacrificing, well, sophistication.
</p>
<p>
Another important principle is to keep the concepts orthogonal.
Methods can be implemented for any type; structures represent data while
interfaces represent abstraction; and so on. Orthogonality makes it
easier to understand what happens when things combine.
</p>
<h2 id="Usage">Usage</h2>
<h3 id="internal_usage">
Is Google using Go internally?</h3>
<p>
Yes. Go is used widely in production inside Google.
One easy example is the server behind
<a href="//golang.org">golang.org</a>.
It's just the <a href="/cmd/godoc"><code>godoc</code></a>
document server running in a production configuration on
<a href="https://developers.google.com/appengine/">Google App Engine</a>.
</p>
<p>
A more significant instance is Google's download server, <code>dl.google.com</code>,
which delivers Chrome binaries and other large installables such as <code>apt-get</code>
packages.
</p>
<p>
Go is not the only language used at Google, far from it, but it is a key language
for a number of areas including
<a href="https://talks.golang.org/2013/go-sreops.slide">site reliability
engineering (SRE)</a>
and large-scale data processing.
</p>
<h3 id="external_usage">
What other companies use Go?</h3>
<p>
Go usage is growing worldwide, especially but by no means exclusively
in the cloud computing space.
A couple of major cloud infrastructure projects written in Go are
Docker and Kubernetes,
but there are many more.
</p>
<p>
It's not just cloud, though.
The Go Wiki includes a
<a href="https://github.com/golang/go/wiki/GoUsers">page</a>,
updated regularly, that lists some of the many companies using Go.
</p>
<p>
The Wiki also has a page with links to
<a href="https://github.com/golang/go/wiki/SuccessStories">success stories</a>
about companies and projects that are using the language.
</p>
<h3 id="Do_Go_programs_link_with_Cpp_programs">
Do Go programs link with C/C++ programs?</h3>
<p>
It is possible to use C and Go together in the same address space,
but it is not a natural fit and can require special interface software.
Also, linking C with Go code gives up the memory
safety and stack management properties that Go provides.
Sometimes it's absolutely necessary to use C libraries to solve a problem,
but doing so always introduces an element of risk not present with
pure Go code, so do so with care.
</p>
<p>
If you do need to use C with Go, how to proceed depends on the Go
compiler implementation.
There are three Go compiler implementations supported by the
Go team.
These are <code>gc</code>, the default compiler,
<code>gccgo</code>, which uses the GCC back end,
and a somewhat less mature <code>gollvm</code>, which uses the LLVM infrastructure.
</p>
<p>
<code>Gc</code> uses a different calling convention and linker from C and
therefore cannot be called directly from C programs, or vice versa.
The <a href="/cmd/cgo/"><code>cgo</code></a> program provides the mechanism for a
&ldquo;foreign function interface&rdquo; to allow safe calling of
C libraries from Go code.
SWIG extends this capability to C++ libraries.
</p>
<p>
You can also use <code>cgo</code> and SWIG with <code>Gccgo</code> and <code>gollvm</code>.
Since they use a traditional API, it's also possible, with great care,
to link code from these compilers directly with GCC/LLVM-compiled C or C++ programs.
However, doing so safely requires an understanding of the calling conventions for
all languages concerned, as well as concern for stack limits when calling C or C++
from Go.
</p>
<h3 id="ide">
What IDEs does Go support?</h3>
<p>
The Go project does not include a custom IDE, but the language and
libraries have been designed to make it easy to analyze source code.
As a consequence, most well-known editors and IDEs support Go well,
either directly or through a plugin.
</p>
<p>
The list of well-known IDEs and editors that have good Go support
available includes Emacs, Vim, VSCode, Atom, Eclipse, Sublime, IntelliJ
(through a custom variant called Goland), and many more.
Chances are your favorite environment is a productive one for
programming in Go.
</p>
<h3 id="protocol_buffers">
Does Go support Google's protocol buffers?</h3>
<p>
A separate open source project provides the necessary compiler plugin and library.
It is available at
<a href="//github.com/golang/protobuf">github.com/golang/protobuf/</a>.
</p>
<h3 id="Can_I_translate_the_Go_home_page">
Can I translate the Go home page into another language?</h3>
<p>
Absolutely. We encourage developers to make Go Language sites in their own languages.
However, if you choose to add the Google logo or branding to your site
(it does not appear on <a href="//golang.org/">golang.org</a>),
you will need to abide by the guidelines at
<a href="//www.google.com/permissions/guidelines.html">www.google.com/permissions/guidelines.html</a>
</p>
<h2 id="Design">Design</h2>
<h3 id="runtime">
Does Go have a runtime?</h3>
<p>
Go does have an extensive library, called the <em>runtime</em>,
that is part of every Go program.
The runtime library implements garbage collection, concurrency,
stack management, and other critical features of the Go language.
Although it is more central to the language, Go's runtime is analogous
to <code>libc</code>, the C library.
</p>
<p>
It is important to understand, however, that Go's runtime does not
include a virtual machine, such as is provided by the Java runtime.
Go programs are compiled ahead of time to native machine code
(or JavaScript or WebAssembly, for some variant implementations).
Thus, although the term is often used to describe the virtual
environment in which a program runs, in Go the word &ldquo;runtime&rdquo;
is just the name given to the library providing critical language services.
</p>
<h3 id="unicode_identifiers">
What's up with Unicode identifiers?</h3>
<p>
When designing Go, we wanted to make sure that it was not
overly ASCII-centric,
which meant extending the space of identifiers from the
confines of 7-bit ASCII.
Go's rule&mdash;identifier characters must be
letters or digits as defined by Unicode&mdash;is simple to understand
and to implement but has restrictions.
Combining characters are
excluded by design, for instance,
and that excludes some languages such as Devanagari.
</p>
<p>
This rule has one other unfortunate consequence.
Since an exported identifier must begin with an
upper-case letter, identifiers created from characters
in some languages can, by definition, not be exported.
For now the
only solution is to use something like <code>X日本語</code>, which
is clearly unsatisfactory.
</p>
<p>
Since the earliest version of the language, there has been considerable
thought into how best to expand the identifier space to accommodate
programmers using other native languages.
Exactly what to do remains an active topic of discussion, and a future
version of the language may be more liberal in its definition
of an identifier.
For instance, it might adopt some of the ideas from the Unicode
organization's <a href="http://unicode.org/reports/tr31/">recommendations</a>
for identifiers.
Whatever happens, it must be done compatibly while preserving
(or perhaps expanding) the way letter case determines visibility of
identifiers, which remains one of our favorite features of Go.
</p>
<p>
For the time being, we have a simple rule that can be expanded later
without breaking programs, one that avoids bugs that would surely arise
from a rule that admits ambiguous identifiers.
</p>
<h3 id="Why_doesnt_Go_have_feature_X">Why does Go not have feature X?</h3>
<p>
Every language contains novel features and omits someone's favorite
feature. Go was designed with an eye on felicity of programming, speed of
compilation, orthogonality of concepts, and the need to support features
such as concurrency and garbage collection. Your favorite feature may be
missing because it doesn't fit, because it affects compilation speed or
clarity of design, or because it would make the fundamental system model
too difficult.
</p>
<p>
If it bothers you that Go is missing feature <var>X</var>,
please forgive us and investigate the features that Go does have. You might find that
they compensate in interesting ways for the lack of <var>X</var>.
</p>
<h3 id="generics">
Why does Go not have generic types?</h3>
<p>
Generics may well be added at some point. We don't feel an urgency for
them, although we understand some programmers do.
</p>
<p>
Go was intended as a language for writing server programs that would be
easy to maintain over time.
(See <a href="https://talks.golang.org/2012/splash.article">this
article</a> for more background.)
The design concentrated on things like scalability, readability, and
concurrency.
Polymorphic programming did not seem essential to the language's
goals at the time, and so was left out for simplicity.
</p>
<p>
The language is more mature now, and there is scope to consider
some form of generic programming.
However, there remain some caveats.
</p>
<p>
Generics are convenient but they come at a cost in
complexity in the type system and run-time. We haven't yet found a
design that gives value proportionate to the complexity, although we
continue to think about it. Meanwhile, Go's built-in maps and slices,
plus the ability to use the empty interface to construct containers
(with explicit unboxing) mean in many cases it is possible to write
code that does what generics would enable, if less smoothly.
</p>
<p>
The topic remains open.
For a look at several previous unsuccessful attempts to
design a good generics solution for Go, see
<a href="https://golang.org/issue/15292">this proposal</a>.
</p>
<h3 id="exceptions">
Why does Go not have exceptions?</h3>
<p>
We believe that coupling exceptions to a control
structure, as in the <code>try-catch-finally</code> idiom, results in
convoluted code. It also tends to encourage programmers to label
too many ordinary errors, such as failing to open a file, as
exceptional.
</p>
<p>
Go takes a different approach. For plain error handling, Go's multi-value
returns make it easy to report an error without overloading the return value.
<a href="/doc/articles/error_handling.html">A canonical error type, coupled
with Go's other features</a>, makes error handling pleasant but quite different
from that in other languages.
</p>
<p>
Go also has a couple
of built-in functions to signal and recover from truly exceptional
conditions. The recovery mechanism is executed only as part of a
function's state being torn down after an error, which is sufficient
to handle catastrophe but requires no extra control structures and,
when used well, can result in clean error-handling code.
</p>
<p>
See the <a href="/doc/articles/defer_panic_recover.html">Defer, Panic, and Recover</a> article for details.
Also, the <a href="https://blog.golang.org/errors-are-values">Errors are values</a> blog post
describes one approach to handling errors cleanly in Go by demonstrating that,
since errors are just values, the full power of Go can deployed in error handling.
</p>
<h3 id="assertions">
Why does Go not have assertions?</h3>
<p>
Go doesn't provide assertions. They are undeniably convenient, but our
experience has been that programmers use them as a crutch to avoid thinking
about proper error handling and reporting. Proper error handling means that
servers continue to operate instead of crashing after a non-fatal error.
Proper error reporting means that errors are direct and to the point,
saving the programmer from interpreting a large crash trace. Precise
errors are particularly important when the programmer seeing the errors is
not familiar with the code.
</p>
<p>
We understand that this is a point of contention. There are many things in
the Go language and libraries that differ from modern practices, simply
because we feel it's sometimes worth trying a different approach.
</p>
<h3 id="csp">
Why build concurrency on the ideas of CSP?</h3>
<p>
Concurrency and multi-threaded programming have over time
developed a reputation for difficulty. We believe this is due partly to complex
designs such as
<a href="https://en.wikipedia.org/wiki/POSIX_Threads">pthreads</a>
and partly to overemphasis on low-level details
such as mutexes, condition variables, and memory barriers.
Higher-level interfaces enable much simpler code, even if there are still
mutexes and such under the covers.
</p>
<p>
One of the most successful models for providing high-level linguistic support
for concurrency comes from Hoare's Communicating Sequential Processes, or CSP.
Occam and Erlang are two well known languages that stem from CSP.
Go's concurrency primitives derive from a different part of the family tree
whose main contribution is the powerful notion of channels as first class objects.
Experience with several earlier languages has shown that the CSP model
fits well into a procedural language framework.
</p>
<h3 id="goroutines">
Why goroutines instead of threads?</h3>
<p>
Goroutines are part of making concurrency easy to use. The idea, which has
been around for a while, is to multiplex independently executing
functions&mdash;coroutines&mdash;onto a set of threads.
When a coroutine blocks, such as by calling a blocking system call,
the run-time automatically moves other coroutines on the same operating
system thread to a different, runnable thread so they won't be blocked.
The programmer sees none of this, which is the point.
The result, which we call goroutines, can be very cheap: they have little
overhead beyond the memory for the stack, which is just a few kilobytes.
</p>
<p>
To make the stacks small, Go's run-time uses resizable, bounded stacks. A newly
minted goroutine is given a few kilobytes, which is almost always enough.
When it isn't, the run-time grows (and shrinks) the memory for storing
the stack automatically, allowing many goroutines to live in a modest
amount of memory.
The CPU overhead averages about three cheap instructions per function call.
It is practical to create hundreds of thousands of goroutines in the same
address space.
If goroutines were just threads, system resources would
run out at a much smaller number.
</p>
<h3 id="atomic_maps">
Why are map operations not defined to be atomic?</h3>
<p>
After long discussion it was decided that the typical use of maps did not require
safe access from multiple goroutines, and in those cases where it did, the map was
probably part of some larger data structure or computation that was already
synchronized. Therefore requiring that all map operations grab a mutex would slow
down most programs and add safety to few. This was not an easy decision,
however, since it means uncontrolled map access can crash the program.
</p>
<p>
The language does not preclude atomic map updates. When required, such
as when hosting an untrusted program, the implementation could interlock
map access.
</p>
<p>
Map access is unsafe only when updates are occurring.
As long as all goroutines are only reading—looking up elements in the map,
including iterating through it using a
<code>for</code> <code>range</code> loop—and not changing the map
by assigning to elements or doing deletions,
it is safe for them to access the map concurrently without synchronization.
</p>
<p>
As an aid to correct map use, some implementations of the language
contain a special check that automatically reports at run time when a map is modified
unsafely by concurrent execution.
</p>
<h3 id="language_changes">
Will you accept my language change?</h3>
<p>
People often suggest improvements to the language—the
<a href="//groups.google.com/group/golang-nuts">mailing list</a>
contains a rich history of such discussions—but very few of these changes have
been accepted.
</p>
<p>
Although Go is an open source project, the language and libraries are protected
by a <a href="/doc/go1compat.html">compatibility promise</a> that prevents
changes that break existing programs, at least at the source code level
(programs may need to be recompiled occasionally to stay current).
If your proposal violates the Go 1 specification we cannot even entertain the
idea, regardless of its merit.
A future major release of Go may be incompatible with Go 1, but discussions
on that topic have only just begun and one thing is certain:
there will be very few such incompatibilities introduced in the process.
Moreover, the compatibility promise encourages us to provide an automatic path
forward for old programs to adapt should that situation arise.
</p>
<p>
Even if your proposal is compatible with the Go 1 spec, it might
not be in the spirit of Go's design goals.
The article <i><a href="//talks.golang.org/2012/splash.article">Go
at Google: Language Design in the Service of Software Engineering</a></i>
explains Go's origins and the motivation behind its design.
</p>
<h2 id="types">Types</h2>
<h3 id="Is_Go_an_object-oriented_language">
Is Go an object-oriented language?</h3>
<p>
Yes and no. Although Go has types and methods and allows an
object-oriented style of programming, there is no type hierarchy.
The concept of &ldquo;interface&rdquo; in Go provides a different approach that
we believe is easy to use and in some ways more general. There are
also ways to embed types in other types to provide something
analogous&mdash;but not identical&mdash;to subclassing.
Moreover, methods in Go are more general than in C++ or Java:
they can be defined for any sort of data, even built-in types such
as plain, &ldquo;unboxed&rdquo; integers.
They are not restricted to structs (classes).
</p>
<p>
Also, the lack of a type hierarchy makes &ldquo;objects&rdquo; in Go feel much more
lightweight than in languages such as C++ or Java.
</p>
<h3 id="How_do_I_get_dynamic_dispatch_of_methods">
How do I get dynamic dispatch of methods?</h3>
<p>
The only way to have dynamically dispatched methods is through an
interface. Methods on a struct or any other concrete type are always resolved statically.
</p>
<h3 id="inheritance">
Why is there no type inheritance?</h3>
<p>
Object-oriented programming, at least in the best-known languages,
involves too much discussion of the relationships between types,
relationships that often could be derived automatically. Go takes a
different approach.
</p>
<p>
Rather than requiring the programmer to declare ahead of time that two
types are related, in Go a type automatically satisfies any interface
that specifies a subset of its methods. Besides reducing the
bookkeeping, this approach has real advantages. Types can satisfy
many interfaces at once, without the complexities of traditional
multiple inheritance.
Interfaces can be very lightweight&mdash;an interface with
one or even zero methods can express a useful concept.
Interfaces can be added after the fact if a new idea comes along
or for testing&mdash;without annotating the original types.
Because there are no explicit relationships between types
and interfaces, there is no type hierarchy to manage or discuss.
</p>
<p>
It's possible to use these ideas to construct something analogous to
type-safe Unix pipes. For instance, see how <code>fmt.Fprintf</code>
enables formatted printing to any output, not just a file, or how the
<code>bufio</code> package can be completely separate from file I/O,
or how the <code>image</code> packages generate compressed
image files. All these ideas stem from a single interface
(<code>io.Writer</code>) representing a single method
(<code>Write</code>). And that's only scratching the surface.
Go's interfaces have a profound influence on how programs are structured.
</p>
<p>
It takes some getting used to but this implicit style of type
dependency is one of the most productive things about Go.
</p>
<h3 id="methods_on_basics">
Why is <code>len</code> a function and not a method?</h3>
<p>
We debated this issue but decided
implementing <code>len</code> and friends as functions was fine in practice and
didn't complicate questions about the interface (in the Go type sense)
of basic types.
</p>
<h3 id="overloading">
Why does Go not support overloading of methods and operators?</h3>
<p>
Method dispatch is simplified if it doesn't need to do type matching as well.
Experience with other languages told us that having a variety of
methods with the same name but different signatures was occasionally useful
but that it could also be confusing and fragile in practice. Matching only by name
and requiring consistency in the types was a major simplifying decision
in Go's type system.
</p>
<p>
Regarding operator overloading, it seems more a convenience than an absolute
requirement. Again, things are simpler without it.
</p>
<h3 id="implements_interface">
Why doesn't Go have "implements" declarations?</h3>
<p>
A Go type satisfies an interface by implementing the methods of that interface,
nothing more. This property allows interfaces to be defined and used without
needing to modify existing code. It enables a kind of
<a href="https://en.wikipedia.org/wiki/Structural_type_system">structural typing</a> that
promotes separation of concerns and improves code re-use, and makes it easier
to build on patterns that emerge as the code develops.
The semantics of interfaces is one of the main reasons for Go's nimble,
lightweight feel.
</p>
<p>
See the <a href="#inheritance">question on type inheritance</a> for more detail.
</p>
<h3 id="guarantee_satisfies_interface">
How can I guarantee my type satisfies an interface?</h3>
<p>
You can ask the compiler to check that the type <code>T</code> implements the
interface <code>I</code> by attempting an assignment using the zero value for
<code>T</code> or pointer to <code>T</code>, as appropriate:
</p>
<pre>
type T struct{}
var _ I = T{} // Verify that T implements I.
var _ I = (*T)(nil) // Verify that *T implements I.
</pre>
<p>
If <code>T</code> (or <code>*T</code>, accordingly) doesn't implement
<code>I</code>, the mistake will be caught at compile time.
</p>
<p>
If you wish the users of an interface to explicitly declare that they implement
it, you can add a method with a descriptive name to the interface's method set.
For example:
</p>
<pre>
type Fooer interface {
Foo()
ImplementsFooer()
}
</pre>
<p>
A type must then implement the <code>ImplementsFooer</code> method to be a
<code>Fooer</code>, clearly documenting the fact and announcing it in
<a href="/cmd/godoc/">godoc</a>'s output.
</p>
<pre>
type Bar struct{}
func (b Bar) ImplementsFooer() {}
func (b Bar) Foo() {}
</pre>
<p>
Most code doesn't make use of such constraints, since they limit the utility of
the interface idea. Sometimes, though, they're necessary to resolve ambiguities
among similar interfaces.
</p>
<h3 id="t_and_equal_interface">
Why doesn't type T satisfy the Equal interface?</h3>
<p>
Consider this simple interface to represent an object that can compare
itself with another value:
</p>
<pre>
type Equaler interface {
Equal(Equaler) bool
}
</pre>
<p>
and this type, <code>T</code>:
</p>
<pre>
type T int
func (t T) Equal(u T) bool { return t == u } // does not satisfy Equaler
</pre>
<p>
Unlike the analogous situation in some polymorphic type systems,
<code>T</code> does not implement <code>Equaler</code>.
The argument type of <code>T.Equal</code> is <code>T</code>,
not literally the required type <code>Equaler</code>.
</p>
<p>
In Go, the type system does not promote the argument of
<code>Equal</code>; that is the programmer's responsibility, as
illustrated by the type <code>T2</code>, which does implement
<code>Equaler</code>:
</p>
<pre>
type T2 int
func (t T2) Equal(u Equaler) bool { return t == u.(T2) } // satisfies Equaler
</pre>
<p>
Even this isn't like other type systems, though, because in Go <em>any</em>
type that satisfies <code>Equaler</code> could be passed as the
argument to <code>T2.Equal</code>, and at run time we must
check that the argument is of type <code>T2</code>.
Some languages arrange to make that guarantee at compile time.
</p>
<p>
A related example goes the other way:
</p>
<pre>
type Opener interface {
Open() Reader
}
func (t T3) Open() *os.File
</pre>
<p>
In Go, <code>T3</code> does not satisfy <code>Opener</code>,
although it might in another language.
</p>
<p>
While it is true that Go's type system does less for the programmer
in such cases, the lack of subtyping makes the rules about
interface satisfaction very easy to state: are the function's names
and signatures exactly those of the interface?
Go's rule is also easy to implement efficiently.
We feel these benefits offset the lack of
automatic type promotion. Should Go one day adopt some form of polymorphic
typing, we expect there would be a way to express the idea of these
examples and also have them be statically checked.
</p>
<h3 id="convert_slice_of_interface">
Can I convert a []T to an []interface{}?</h3>
<p>
Not directly.
It is disallowed by the language specification because the two types
do not have the same representation in memory.
It is necessary to copy the elements individually to the destination
slice. This example converts a slice of <code>int</code> to a slice of
<code>interface{}</code>:
</p>
<pre>
t := []int{1, 2, 3, 4}
s := make([]interface{}, len(t))
for i, v := range t {
s[i] = v
}
</pre>
<h3 id="convert_slice_with_same_underlying_type">
Can I convert []T1 to []T2 if T1 and T2 have the same underlying type?</h3>
This last line of this code sample does not compile.
<pre>
type T1 int
type T2 int
var t1 T1
var x = T2(t1) // OK
var st1 []T1
var sx = ([]T2)(st1) // NOT OK
</pre>
<p>
In Go, types are closely tied to methods, in that every named type has
a (possibly empty) method set.
The general rule is that you can change the name of the type being
converted (and thus possibly change its method set) but you can't
change the name (and method set) of elements of a composite type.
Go requires you to be explicit about type conversions.
</p>
<h3 id="nil_error">
Why is my nil error value not equal to nil?
</h3>
<p>
Under the covers, interfaces are implemented as two elements, a type <code>T</code>
and a value <code>V</code>.
<code>V</code> is a concrete value such as an <code>int</code>,
<code>struct</code> or pointer, never an interface itself, and has
type <code>T</code>.
For instance, if we store the <code>int</code> value 3 in an interface,
the resulting interface value has, schematically,
(<code>T=int</code>, <code>V=3</code>).
The value <code>V</code> is also known as the interface's
<em>dynamic</em> value,
since a given interface variable might hold different values <code>V</code>
(and corresponding types <code>T</code>)
during the execution of the program.
</p>
<p>
An interface value is <code>nil</code> only if the <code>V</code> and <code>T</code>
are both unset, (<code>T=nil</code>, <code>V</code> is not set),
In particular, a <code>nil</code> interface will always hold a <code>nil</code> type.
If we store a <code>nil</code> pointer of type <code>*int</code> inside
an interface value, the inner type will be <code>*int</code> regardless of the value of the pointer:
(<code>T=*int</code>, <code>V=nil</code>).
Such an interface value will therefore be non-<code>nil</code>
<em>even when the pointer value <code>V</code> inside is</em> <code>nil</code>.
</p>
<p>
This situation can be confusing, and arises when a <code>nil</code> value is
stored inside an interface value such as an <code>error</code> return:
</p>
<pre>
func returnsError() error {
var p *MyError = nil
if bad() {
p = ErrBad
}
return p // Will always return a non-nil error.
}
</pre>
<p>
If all goes well, the function returns a <code>nil</code> <code>p</code>,
so the return value is an <code>error</code> interface
value holding (<code>T=*MyError</code>, <code>V=nil</code>).
This means that if the caller compares the returned error to <code>nil</code>,
it will always look as if there was an error even if nothing bad happened.
To return a proper <code>nil</code> <code>error</code> to the caller,
the function must return an explicit <code>nil</code>:
</p>
<pre>
func returnsError() error {
if bad() {
return ErrBad
}
return nil
}
</pre>
<p>
It's a good idea for functions
that return errors always to use the <code>error</code> type in
their signature (as we did above) rather than a concrete type such
as <code>*MyError</code>, to help guarantee the error is
created correctly. As an example,
<a href="/pkg/os/#Open"><code>os.Open</code></a>
returns an <code>error</code> even though, if not <code>nil</code>,
it's always of concrete type
<a href="/pkg/os/#PathError"><code>*os.PathError</code></a>.
</p>
<p>
Similar situations to those described here can arise whenever interfaces are used.
Just keep in mind that if any concrete value
has been stored in the interface, the interface will not be <code>nil</code>.
For more information, see
<a href="/doc/articles/laws_of_reflection.html">The Laws of Reflection</a>.
</p>
<h3 id="unions">
Why are there no untagged unions, as in C?</h3>
<p>
Untagged unions would violate Go's memory safety
guarantees.
</p>
<h3 id="variant_types">
Why does Go not have variant types?</h3>
<p>
Variant types, also known as algebraic types, provide a way to specify
that a value might take one of a set of other types, but only those
types. A common example in systems programming would specify that an
error is, say, a network error, a security error or an application
error and allow the caller to discriminate the source of the problem
by examining the type of the error. Another example is a syntax tree
in which each node can be a different type: declaration, statement,
assignment and so on.
</p>
<p>
We considered adding variant types to Go, but after discussion
decided to leave them out because they overlap in confusing ways
with interfaces. What would happen if the elements of a variant type
were themselves interfaces?
</p>
<p>
Also, some of what variant types address is already covered by the
language. The error example is easy to express using an interface
value to hold the error and a type switch to discriminate cases. The
syntax tree example is also doable, although not as elegantly.
</p>
<h3 id="covariant_types">
Why does Go not have covariant result types?</h3>
<p>
Covariant result types would mean that an interface like
</p>
<pre>
type Copyable interface {
Copy() interface{}
}
</pre>
<p>
would be satisfied by the method
</p>
<pre>
func (v Value) Copy() Value
</pre>
<p>because <code>Value</code> implements the empty interface.
In Go method types must match exactly, so <code>Value</code> does not
implement <code>Copyable</code>.
Go separates the notion of what a
type does&mdash;its methods&mdash;from the type's implementation.
If two methods return different types, they are not doing the same thing.
Programmers who want covariant result types are often trying to
express a type hierarchy through interfaces.
In Go it's more natural to have a clean separation between interface
and implementation.
</p>
<h2 id="values">Values</h2>
<h3 id="conversions">
Why does Go not provide implicit numeric conversions?</h3>
<p>
The convenience of automatic conversion between numeric types in C is
outweighed by the confusion it causes. When is an expression unsigned?
How big is the value? Does it overflow? Is the result portable, independent
of the machine on which it executes?
It also complicates the compiler; &ldquo;the usual arithmetic conversions&rdquo;
are not easy to implement and inconsistent across architectures.
For reasons of portability, we decided to make things clear and straightforward
at the cost of some explicit conversions in the code.
The definition of constants in Go&mdash;arbitrary precision values free
of signedness and size annotations&mdash;ameliorates matters considerably,
though.
</p>
<p>
A related detail is that, unlike in C, <code>int</code> and <code>int64</code>
are distinct types even if <code>int</code> is a 64-bit type. The <code>int</code>
type is generic; if you care about how many bits an integer holds, Go
encourages you to be explicit.
</p>
<h3 id="constants">
How do constants work in Go?</h3>
<p>
Although Go is strict about conversion between variables of different
numeric types, constants in the language are much more flexible.
Literal constants such as <code>23</code>, <code>3.14159</code>
and <a href="/pkg/math/#pkg-constants"><code>math.Pi</code></a>
occupy a sort of ideal number space, with arbitrary precision and
no overflow or underflow.
For instance, the value of <code>math.Pi</code> is specified to 63 places
in the source code, and constant expressions involving the value keep
precision beyond what a <code>float64</code> could hold.
Only when the constant or constant expression is assigned to a
variable&mdash;a memory location in the program&mdash;does
it become a "computer" number with
the usual floating-point properties and precision.
</p>
<p>
Also,
because they are just numbers, not typed values, constants in Go can be
used more freely than variables, thereby softening some of the awkwardness
around the strict conversion rules.
One can write expressions such as
</p>
<pre>
sqrt2 := math.Sqrt(2)
</pre>
<p>
without complaint from the compiler because the ideal number <code>2</code>
can be converted safely and accurately
to a <code>float64</code> for the call to <code>math.Sqrt</code>.
</p>
<p>
A blog post titled <a href="https://blog.golang.org/constants">Constants</a>
explores this topic in more detail.
</p>
<h3 id="builtin_maps">
Why are maps built in?</h3>
<p>
The same reason strings are: they are such a powerful and important data
structure that providing one excellent implementation with syntactic support
makes programming more pleasant. We believe that Go's implementation of maps
is strong enough that it will serve for the vast majority of uses.
If a specific application can benefit from a custom implementation, it's possible
to write one but it will not be as convenient syntactically; this seems a reasonable tradeoff.
</p>
<h3 id="map_keys">
Why don't maps allow slices as keys?</h3>
<p>
Map lookup requires an equality operator, which slices do not implement.
They don't implement equality because equality is not well defined on such types;
there are multiple considerations involving shallow vs. deep comparison, pointer vs.
value comparison, how to deal with recursive types, and so on.
We may revisit this issue&mdash;and implementing equality for slices
will not invalidate any existing programs&mdash;but without a clear idea of what
equality of slices should mean, it was simpler to leave it out for now.
</p>
<p>
In Go 1, unlike prior releases, equality is defined for structs and arrays, so such
types can be used as map keys. Slices still do not have a definition of equality, though.
</p>
<h3 id="references">
Why are maps, slices, and channels references while arrays are values?</h3>
<p>
There's a lot of history on that topic. Early on, maps and channels
were syntactically pointers and it was impossible to declare or use a
non-pointer instance. Also, we struggled with how arrays should work.
Eventually we decided that the strict separation of pointers and
values made the language harder to use. Changing these
types to act as references to the associated, shared data structures resolved
these issues. This change added some regrettable complexity to the
language but had a large effect on usability: Go became a more
productive, comfortable language when it was introduced.
</p>
<h2 id="Writing_Code">Writing Code</h2>
<h3 id="How_are_libraries_documented">
How are libraries documented?</h3>
<p>
There is a program, <code>godoc</code>, written in Go, that extracts
package documentation from the source code and serves it as a web
page with links to declarations, files, and so on.
An instance is running at
<a href="/pkg/">golang.org/pkg/</a>.
In fact, <code>godoc</code> implements the full site at
<a href="/">golang.org/</a>.
</p>
<p>
A <code>godoc</code> instance may be configured to provide rich,
interactive static analyses of symbols in the programs it displays; details are
listed <a href="https://golang.org/lib/godoc/analysis/help.html">here</a>.
</p>
<p>
For access to documentation from the command line, the
<a href="https://golang.org/pkg/cmd/go/">go</a> tool has a
<a href="https://golang.org/pkg/cmd/go/#hdr-Show_documentation_for_package_or_symbol">doc</a>
subcommand that provides a textual interface to the same information.
</p>
<h3 id="Is_there_a_Go_programming_style_guide">
Is there a Go programming style guide?</h3>
<p>
There is no explicit style guide, although there is certainly
a recognizable "Go style".
</p>
<p>
Go has established conventions to guide decisions around
naming, layout, and file organization.
The document <a href="effective_go.html">Effective Go</a>
contains some advice on these topics.
More directly, the program <code>gofmt</code> is a pretty-printer
whose purpose is to enforce layout rules; it replaces the usual
compendium of do's and don'ts that allows interpretation.
All the Go code in the repository, and the vast majority in the
open source world, has been run through <code>gofmt</code>.
</p>
<p>
The document titled
<a href="//golang.org/s/comments">Go Code Review Comments</a>
is a collection of very short essays about details of Go idiom that are often
missed by programmers.
It is a handy reference for people doing code reviews for Go projects.
</p>
<h3 id="How_do_I_submit_patches_to_the_Go_libraries">
How do I submit patches to the Go libraries?</h3>
<p>
The library sources are in the <code>src</code> directory of the repository.
If you want to make a significant change, please discuss on the mailing list before embarking.
</p>
<p>
See the document
<a href="contribute.html">Contributing to the Go project</a>
for more information about how to proceed.
</p>
<h3 id="git_https">
Why does "go get" use HTTPS when cloning a repository?</h3>
<p>
Companies often permit outgoing traffic only on the standard TCP ports 80 (HTTP)
and 443 (HTTPS), blocking outgoing traffic on other ports, including TCP port 9418
(git) and TCP port 22 (SSH).
When using HTTPS instead of HTTP, <code>git</code> enforces certificate validation by
default, providing protection against man-in-the-middle, eavesdropping and tampering attacks.
The <code>go get</code> command therefore uses HTTPS for safety.
</p>
<p>
<code>Git</code> can be configured to authenticate over HTTPS or to use SSH in place of HTTPS.
To authenticate over HTTPS, you can add a line
to the <code>$HOME/.netrc</code> file that git consults:
</p>
<pre>
machine github.com login <i>USERNAME</i> password <i>APIKEY</i>
</pre>
<p>
For GitHub accounts, the password can be a
<a href="https://help.github.com/articles/creating-a-personal-access-token-for-the-command-line/">personal access token</a>.
</p>
<p>
<code>Git</code> can also be configured to use SSH in place of HTTPS for URLs matching a given prefix.
For example, to use SSH for all GitHub access,
add these lines to your <code>~/.gitconfig</code>:
</p>
<pre>
[url "ssh://git@github.com/"]
insteadOf = https://github.com/
</pre>
<h3 id="get_version">
How should I manage package versions using "go get"?</h3>
<p>
Since the inception of the project, Go has had no explicit concept of package versions,
but that is changing.
Versioning is a source of significant complexity, especially in large code bases,
and it has taken some time to develop an
approach that works well at scale in a large enough
variety of situations to be appropriate to supply to all Go users.
</p>
<p>
The Go 1.11 release adds new, experimental support
for package versioning to the <code>go</code> command,
in the form of Go modules.
For more information, see the <a href="/doc/go1.11#modules">Go 1.11 release notes</a>
and the <a href="/cmd/go#hdr-Modules__module_versions__and_more"><code>go</code> command documentation</a>.
</p>
<p>
Regardless of the actual package management technology,
"go get" and the larger Go toolchain does provide isolation of
packages with different import paths.
For example, the standard library's <code>html/template</code> and <code>text/template</code>
coexist even though both are "package template".
This observation leads to some advice for package authors and package users.
</p>
<p>
Packages intended for public use should try to maintain backwards compatibility as they evolve.
The <a href="/doc/go1compat.html">Go 1 compatibility guidelines</a> are a good reference here:
don't remove exported names, encourage tagged composite literals, and so on.
If different functionality is required, add a new name instead of changing an old one.
If a complete break is required, create a new package with a new import path.
</p>
<p>
If you're using an externally supplied package and worry that it might change in
unexpected ways, but are not yet using Go modules,
the simplest solution is to copy it to your local repository.
This is the approach Google takes internally and is supported by the
<code>go</code> command through a technique called "vendoring".
This involves
storing a copy of the dependency under a new import path that identifies it as a local copy.
See the <a href="https://golang.org/s/go15vendor">design
document</a> for details.
</p>
<h2 id="Pointers">Pointers and Allocation</h2>
<h3 id="pass_by_value">
When are function parameters passed by value?</h3>
<p>
As in all languages in the C family, everything in Go is passed by value.
That is, a function always gets a copy of the
thing being passed, as if there were an assignment statement assigning the
value to the parameter. For instance, passing an <code>int</code> value
to a function makes a copy of the <code>int</code>, and passing a pointer
value makes a copy of the pointer, but not the data it points to.
(See a <a href="/doc/faq#methods_on_values_or_pointers">later
section</a> for a discussion of how this affects method receivers.)
</p>
<p>
Map and slice values behave like pointers: they are descriptors that
contain pointers to the underlying map or slice data. Copying a map or
slice value doesn't copy the data it points to. Copying an interface value
makes a copy of the thing stored in the interface value. If the interface
value holds a struct, copying the interface value makes a copy of the
struct. If the interface value holds a pointer, copying the interface value
makes a copy of the pointer, but again not the data it points to.
</p>
<p>
Note that this discussion is about the semantics of the operations.
Actual implementations may apply optimizations to avoid copying
as long as the optimizations do not change the semantics.
</p>
<h3 id="pointer_to_interface">
When should I use a pointer to an interface?</h3>
<p>
Almost never. Pointers to interface values arise only in rare, tricky situations involving
disguising an interface value's type for delayed evaluation.
</p>
<p>
It is a common mistake to pass a pointer to an interface value
to a function expecting an interface. The compiler will complain about this
error but the situation can still be confusing, because sometimes a
<a href="#different_method_sets">pointer
is necessary to satisfy an interface</a>.
The insight is that although a pointer to a concrete type can satisfy
an interface, with one exception <em>a pointer to an interface can never satisfy an interface</em>.
</p>
<p>
Consider the variable declaration,
</p>
<pre>
var w io.Writer
</pre>
<p>
The printing function <code>fmt.Fprintf</code> takes as its first argument
a value that satisfies <code>io.Writer</code>—something that implements
the canonical <code>Write</code> method. Thus we can write
</p>
<pre>
fmt.Fprintf(w, "hello, world\n")
</pre>
<p>
If however we pass the address of <code>w</code>, the program will not compile.
</p>
<pre>
fmt.Fprintf(&amp;w, "hello, world\n") // Compile-time error.
</pre>
<p>
The one exception is that any value, even a pointer to an interface, can be assigned to
a variable of empty interface type (<code>interface{}</code>).
Even so, it's almost certainly a mistake if the value is a pointer to an interface;
the result can be confusing.
</p>
<h3 id="methods_on_values_or_pointers">
Should I define methods on values or pointers?</h3>
<pre>
func (s *MyStruct) pointerMethod() { } // method on pointer
func (s MyStruct) valueMethod() { } // method on value
</pre>
<p>
For programmers unaccustomed to pointers, the distinction between these
two examples can be confusing, but the situation is actually very simple.
When defining a method on a type, the receiver (<code>s</code> in the above
examples) behaves exactly as if it were an argument to the method.
Whether to define the receiver as a value or as a pointer is the same
question, then, as whether a function argument should be a value or
a pointer.
There are several considerations.
</p>
<p>
First, and most important, does the method need to modify the
receiver?
If it does, the receiver <em>must</em> be a pointer.
(Slices and maps act as references, so their story is a little
more subtle, but for instance to change the length of a slice
in a method the receiver must still be a pointer.)
In the examples above, if <code>pointerMethod</code> modifies
the fields of <code>s</code>,
the caller will see those changes, but <code>valueMethod</code>
is called with a copy of the caller's argument (that's the definition
of passing a value), so changes it makes will be invisible to the caller.
</p>
<p>
By the way, in Java method receivers are always pointers,
although their pointer nature is somewhat disguised
(and there is a proposal to add value receivers to the language).
It is the value receivers in Go that are unusual.
</p>
<p>
Second is the consideration of efficiency. If the receiver is large,
a big <code>struct</code> for instance, it will be much cheaper to
use a pointer receiver.
</p>
<p>
Next is consistency. If some of the methods of the type must have
pointer receivers, the rest should too, so the method set is
consistent regardless of how the type is used.
See the section on <a href="#different_method_sets">method sets</a>
for details.
</p>
<p>
For types such as basic types, slices, and small <code>structs</code>,
a value receiver is very cheap so unless the semantics of the method
requires a pointer, a value receiver is efficient and clear.
</p>
<h3 id="new_and_make">
What's the difference between new and make?</h3>
<p>
In short: <code>new</code> allocates memory, while <code>make</code> initializes
the slice, map, and channel types.
</p>
<p>
See the <a href="/doc/effective_go.html#allocation_new">relevant section
of Effective Go</a> for more details.
</p>
<h3 id="q_int_sizes">
What is the size of an <code>int</code> on a 64 bit machine?</h3>
<p>
The sizes of <code>int</code> and <code>uint</code> are implementation-specific
but the same as each other on a given platform.
For portability, code that relies on a particular
size of value should use an explicitly sized type, like <code>int64</code>.
On 32-bit machines the compilers use 32-bit integers by default,
while on 64-bit machines integers have 64 bits.
(Historically, this was not always true.)
</p>
<p>
On the other hand, floating-point scalars and complex
types are always sized (there are no <code>float</code> or <code>complex</code> basic types),
because programmers should be aware of precision when using floating-point numbers.
The default type used for an (untyped) floating-point constant is <code>float64</code>.
Thus <code>foo</code> <code>:=</code> <code>3.0</code> declares a variable <code>foo</code>
of type <code>float64</code>.
For a <code>float32</code> variable initialized by an (untyped) constant, the variable type
must be specified explicitly in the variable declaration:
</p>
<pre>
var foo float32 = 3.0
</pre>
<p>
Alternatively, the constant must be given a type with a conversion as in
<code>foo := float32(3.0)</code>.
</p>
<h3 id="stack_or_heap">
How do I know whether a variable is allocated on the heap or the stack?</h3>
<p>
From a correctness standpoint, you don't need to know.
Each variable in Go exists as long as there are references to it.
The storage location chosen by the implementation is irrelevant to the
semantics of the language.
</p>
<p>
The storage location does have an effect on writing efficient programs.
When possible, the Go compilers will allocate variables that are
local to a function in that function's stack frame. However, if the
compiler cannot prove that the variable is not referenced after the
function returns, then the compiler must allocate the variable on the
garbage-collected heap to avoid dangling pointer errors.
Also, if a local variable is very large, it might make more sense
to store it on the heap rather than the stack.
</p>
<p>
In the current compilers, if a variable has its address taken, that variable
is a candidate for allocation on the heap. However, a basic <em>escape
analysis</em> recognizes some cases when such variables will not
live past the return from the function and can reside on the stack.
</p>
<h3 id="Why_does_my_Go_process_use_so_much_virtual_memory">
Why does my Go process use so much virtual memory?</h3>
<p>
The Go memory allocator reserves a large region of virtual memory as an arena
for allocations. This virtual memory is local to the specific Go process; the
reservation does not deprive other processes of memory.
</p>
<p>
To find the amount of actual memory allocated to a Go process, use the Unix
<code>top</code> command and consult the <code>RES</code> (Linux) or
<code>RSIZE</code> (macOS) columns.
<!-- TODO(adg): find out how this works on Windows -->
</p>
<h2 id="Concurrency">Concurrency</h2>
<h3 id="What_operations_are_atomic_What_about_mutexes">
What operations are atomic? What about mutexes?</h3>
<p>
A description of the atomicity of operations in Go can be found in
the <a href="/ref/mem">Go Memory Model</a> document.
</p>
<p>
Low-level synchronization and atomic primitives are available in the
<a href="/pkg/sync">sync</a> and
<a href="/pkg/sync/atomic">sync/atomic</a>
packages.
These packages are good for simple tasks such as incrementing
reference counts or guaranteeing small-scale mutual exclusion.
</p>
<p>
For higher-level operations, such as coordination among
concurrent servers, higher-level techniques can lead
to nicer programs, and Go supports this approach through
its goroutines and channels.
For instance, you can structure your program so that only one
goroutine at a time is ever responsible for a particular piece of data.
That approach is summarized by the original
<a href="https://www.youtube.com/watch?v=PAAkCSZUG1c">Go proverb</a>,
</p>
<p>
Do not communicate by sharing memory. Instead, share memory by communicating.
</p>
<p>
See the <a href="/doc/codewalk/sharemem/">Share Memory By Communicating</a> code walk
and its <a href="https://blog.golang.org/2010/07/share-memory-by-communicating.html">
associated article</a> for a detailed discussion of this concept.
</p>
<p>
Large concurrent programs are likely to borrow from both these toolkits.
</p>
<h3 id="parallel_slow">
Why doesn't my program run faster with more CPUs?</h3>
<p>
Whether a program runs faster with more CPUs depends on the problem
it is solving.
The Go language provides concurrency primitives, such as goroutines
and channels, but concurrency only enables parallelism
when the underlying problem is intrinsically parallel.
Problems that are intrinsically sequential cannot be sped up by adding
more CPUs, while those that can be broken into pieces that can
execute in parallel can be sped up, sometimes dramatically.
</p>
<p>
Sometimes adding more CPUs can slow a program down.
In practical terms, programs that spend more time
synchronizing or communicating than doing useful computation
may experience performance degradation when using
multiple OS threads.
This is because passing data between threads involves switching
contexts, which has significant cost, and that cost can increase
with more CPUs.
For instance, the <a href="/ref/spec#An_example_package">prime sieve example</a>
from the Go specification has no significant parallelism although it launches many
goroutines; increasing the number of threads (CPUs) is more likely to slow it down than
to speed it up.
</p>
<p>
For more detail on this topic see the talk entitled
<a href="//blog.golang.org/2013/01/concurrency-is-not-parallelism.html">Concurrency
is not Parallelism</a>.
<h3 id="number_cpus">
How can I control the number of CPUs?</h3>
<p>
The number of CPUs available simultaneously to executing goroutines is
controlled by the <code>GOMAXPROCS</code> shell environment variable,
whose default value is the number of CPU cores available.
Programs with the potential for parallel execution should therefore
achieve it by default on a multiple-CPU machine.
To change the number of parallel CPUs to use,
set the environment variable or use the similarly-named
<a href="/pkg/runtime/#GOMAXPROCS">function</a>
of the runtime package to configure the
run-time support to utilize a different number of threads.
Setting it to 1 eliminates the possibility of true parallelism,
forcing independent goroutines to take turns executing.
</p>
<p>
The runtime can allocate more threads than the value
of <code>GOMAXPROCS</code> to service multiple outstanding
I/O requests.
<code>GOMAXPROCS</code> only affects how many goroutines
can actually execute at once; arbitrarily more may be blocked
in system calls.
</p>
<p>
Go's goroutine scheduler is not as good as it needs to be, although it
has improved over time.
In the future, it may better optimize its use of OS threads.
For now, if there are performance issues,
setting <code>GOMAXPROCS</code> on a per-application basis may help.
</p>
<h3 id="no_goroutine_id">
Why is there no goroutine ID?</h3>
<p>
Goroutines do not have names; they are just anonymous workers.
They expose no unique identifier, name, or data structure to the programmer.
Some people are surprised by this, expecting the <code>go</code>
statement to return some item that can be used to access and control
the goroutine later.
</p>
<p>
The fundamental reason goroutines are anonymous is so that
the full Go language is available when programming concurrent code.
By contrast, the usage patterns that develop when threads and goroutines are
named can restrict what a library using them can do.
</p>
<p>
Here is an illustration of the difficulties.
Once one names a goroutine and constructs a model around
it, it becomes special, and one is tempted to associate all computation
with that goroutine, ignoring the possibility
of using multiple, possibly shared goroutines for the processing.
If the <code>net/http</code> package associated per-request
state with a goroutine,
clients would be unable to use more goroutines
when serving a request.
</p>
<p>
Moreover, experience with libraries such as those for graphics systems
that require all processing to occur on the "main thread"
has shown how awkward and limiting the approach can be when
deployed in a concurrent language.
The very existence of a special thread or goroutine forces
the programmer to distort the program to avoid crashes
and other problems caused by inadvertently operating
on the wrong thread.
</p>
<p>
For those cases where a particular goroutine is truly special,
the language provides features such as channels that can be
used in flexible ways to interact with it.
</p>
<h2 id="Functions_methods">Functions and Methods</h2>
<h3 id="different_method_sets">
Why do T and *T have different method sets?</h3>
<p>
As the <a href="/ref/spec#Types">Go specification</a> says,
the method set of a type <code>T</code> consists of all methods
with receiver type <code>T</code>,
while that of the corresponding pointer
type <code>*T</code> consists of all methods with receiver <code>*T</code> or
<code>T</code>.
That means the method set of <code>*T</code>
includes that of <code>T</code>),
but not the reverse.
</p>
<p>
This distinction arises because
if an interface value contains a pointer <code>*T</code>,
a method call can obtain a value by dereferencing the pointer,
but if an interface value contains a value <code>T</code>,
there is no safe way for a method call to obtain a pointer.
(Doing so would allow a method to modify the contents of
the value inside the interface, which is not permitted by
the language specification.)
</p>
<p>
Even in cases where the compiler could take the address of a value
to pass to the method, if the method modifies the value the changes
will be lost in the caller.
As an example, if the <code>Write</code> method of
<a href="/pkg/bytes/#Buffer"><code>bytes.Buffer</code></a>
used a value receiver rather than a pointer,
this code:
</p>
<pre>
var buf bytes.Buffer
io.Copy(buf, os.Stdin)
</pre>
<p>
would copy standard input into a <i>copy</i> of <code>buf</code>,
not into <code>buf</code> itself.
This is almost never the desired behavior.
</p>
<h3 id="closures_and_goroutines">
What happens with closures running as goroutines?</h3>
<p>
Some confusion may arise when using closures with concurrency.
Consider the following program:
</p>
<pre>
func main() {
done := make(chan bool)
values := []string{"a", "b", "c"}
for _, v := range values {
go func() {
fmt.Println(v)
done &lt;- true
}()
}
// wait for all goroutines to complete before exiting
for _ = range values {
&lt;-done
}
}
</pre>
<p>
One might mistakenly expect to see <code>a, b, c</code> as the output.
What you'll probably see instead is <code>c, c, c</code>. This is because
each iteration of the loop uses the same instance of the variable <code>v</code>, so
each closure shares that single variable. When the closure runs, it prints the
value of <code>v</code> at the time <code>fmt.Println</code> is executed,
but <code>v</code> may have been modified since the goroutine was launched.
To help detect this and other problems before they happen, run
<a href="/cmd/go/#hdr-Run_go_tool_vet_on_packages"><code>go vet</code></a>.
</p>
<p>
To bind the current value of <code>v</code> to each closure as it is launched, one
must modify the inner loop to create a new variable each iteration.
One way is to pass the variable as an argument to the closure:
</p>
<pre>
for _, v := range values {
go func(<b>u</b> string) {
fmt.Println(<b>u</b>)
done &lt;- true
}(<b>v</b>)
}
</pre>
<p>
In this example, the value of <code>v</code> is passed as an argument to the
anonymous function. That value is then accessible inside the function as
the variable <code>u</code>.
</p>
<p>
Even easier is just to create a new variable, using a declaration style that may
seem odd but works fine in Go:
</p>
<pre>
for _, v := range values {
<b>v := v</b> // create a new 'v'.
go func() {
fmt.Println(<b>v</b>)
done &lt;- true
}()
}
</pre>
<p>
This behavior of the language, not defining a new variable for
each iteration, may have been a mistake in retrospect.
It may be addressed in a later version but, for compatibility,
cannot change in Go version 1.
</p>
<h2 id="Control_flow">Control flow</h2>
<h3 id="Does_Go_have_a_ternary_form">
Why does Go not have the <code>?:</code> operator?</h3>
<p>
There is no ternary testing operation in Go.
You may use the following to achieve the same
result:
</p>
<pre>
if expr {
n = trueVal
} else {
n = falseVal
}
</pre>
<p>
The reason <code>?:</code> is absent from Go is that the language's designers
had seen the operation used too often to create impenetrably complex expressions.
The <code>if-else</code> form, although longer,
is unquestionably clearer.
A language needs only one conditional control flow construct.
</p>
<h2 id="Packages_Testing">Packages and Testing</h2>
<h3 id="How_do_I_create_a_multifile_package">
How do I create a multifile package?</h3>
<p>
Put all the source files for the package in a directory by themselves.
Source files can refer to items from different files at will; there is
no need for forward declarations or a header file.
</p>
<p>
Other than being split into multiple files, the package will compile and test
just like a single-file package.
</p>
<h3 id="How_do_I_write_a_unit_test">
How do I write a unit test?</h3>
<p>
Create a new file ending in <code>_test.go</code> in the same directory
as your package sources. Inside that file, <code>import "testing"</code>
and write functions of the form
</p>
<pre>
func TestFoo(t *testing.T) {
...
}
</pre>
<p>
Run <code>go test</code> in that directory.
That script finds the <code>Test</code> functions,
builds a test binary, and runs it.
</p>
<p>See the <a href="/doc/code.html">How to Write Go Code</a> document,
the <a href="/pkg/testing/"><code>testing</code></a> package
and the <a href="/cmd/go/#hdr-Test_packages"><code>go test</code></a> subcommand for more details.
</p>
<h3 id="testing_framework">
Where is my favorite helper function for testing?</h3>
<p>
Go's standard <a href="/pkg/testing/"><code>testing</code></a> package makes it easy to write unit tests, but it lacks
features provided in other language's testing frameworks such as assertion functions.
An <a href="#assertions">earlier section</a> of this document explained why Go
doesn't have assertions, and
the same arguments apply to the use of <code>assert</code> in tests.
Proper error handling means letting other tests run after one has failed, so
that the person debugging the failure gets a complete picture of what is
wrong. It is more useful for a test to report that
<code>isPrime</code> gives the wrong answer for 2, 3, 5, and 7 (or for
2, 4, 8, and 16) than to report that <code>isPrime</code> gives the wrong
answer for 2 and therefore no more tests were run. The programmer who
triggers the test failure may not be familiar with the code that fails.
Time invested writing a good error message now pays off later when the
test breaks.
</p>
<p>
A related point is that testing frameworks tend to develop into mini-languages
of their own, with conditionals and controls and printing mechanisms,
but Go already has all those capabilities; why recreate them?
We'd rather write tests in Go; it's one fewer language to learn and the
approach keeps the tests straightforward and easy to understand.
</p>
<p>
If the amount of extra code required to write
good errors seems repetitive and overwhelming, the test might work better if
table-driven, iterating over a list of inputs and outputs defined
in a data structure (Go has excellent support for data structure literals).
The work to write a good test and good error messages will then be amortized over many
test cases. The standard Go library is full of illustrative examples, such as in
<a href="/src/fmt/fmt_test.go">the formatting tests for the <code>fmt</code> package</a>.
</p>
<h3 id="x_in_std">
Why isn't <i>X</i> in the standard library?</h3>
<p>
The standard library's purpose is to support the runtime, connect to
the operating system, and provide key functionality that many Go
programs require, such as formatted I/O and networking.
It also contains elements important for web programming, including
cryptography and support for standards like HTTP, JSON, and XML.
</p>
<p>
There is no clear criterion that defines what is included because for
a long time, this was the <i>only</i> Go library.
There are criteria that define what gets added today, however.
</p>
<p>
New additions to the standard library are rare and the bar for
inclusion is high.
Code included in the standard library bears a large ongoing maintenance cost
(often borne by those other than the original author),
is subject to the <a href="/doc/go1compat.html">Go 1 compatibility promise</a>
(blocking fixes to any flaws in the API),
and is subject to the Go
<a href="https://golang.org/s/releasesched">release schedule</a>,
preventing bug fixes from being available to users quickly.
</p>
<p>
Most new code should live outside of the standard library and be accessible
via the <a href="/cmd/go/"><code>go</code> tool</a>'s
<code>go get</code> command.
Such code can have its own maintainers, release cycle,
and compatibility guarantees.
Users can find packages and read their documentation at
<a href="https://godoc.org/">godoc.org</a>.
</p>
<p>
Although there are pieces in the standard library that don't really belong,
such as <code>log/syslog</code>, we continue to maintain everything in the
library because of the Go 1 compatibility promise.
But we encourage most new code to live elsewhere.
</p>
<h2 id="Implementation">Implementation</h2>
<h3 id="What_compiler_technology_is_used_to_build_the_compilers">
What compiler technology is used to build the compilers?</h3>
<p>
There are several production compilers for Go, and a number of others
in development for various platforms.
</p>
<p>
The default compiler, <code>gc</code>, is included with the
Go distribution as part of the support for the <code>go</code>
command.
<code>Gc</code> was originally written in C
because of the difficulties of bootstrapping&mdash;you'd need a Go compiler to
set up a Go environment.
But things have advanced and since the Go 1.5 release the compiler has been
a Go program.
The compiler was converted from C to Go using automatic translation tools, as
described in this <a href="/s/go13compiler">design document</a>
and <a href="https://talks.golang.org/2015/gogo.slide#1">talk</a>.
Thus the compiler is now "self-hosting", which means we needed to face
the bootstrapping problem.
The solution is to have a working Go installation already in place,
just as one normally has with a working C installation.
The story of how to bring up a new Go environment from source
is described <a href="/s/go15bootstrap">here</a> and
<a href="/doc/install/source">here</a>.
</p>
<p>
<code>Gc</code> is written in Go with a recursive descent parser
and uses a custom loader, also written in Go but
based on the Plan 9 loader, to generate ELF/Mach-O/PE binaries.
</p>
<p>
At the beginning of the project we considered using LLVM for
<code>gc</code> but decided it was too large and slow to meet
our performance goals.
More important in retrospect, starting with LLVM would have made it
harder to introduce some of the ABI and related changes, such as
stack management, that Go requires but not are not part of the
standard C setup.
A new <a href="https://go.googlesource.com/gollvm/">LLVM implementation</a>
is starting to come together now, however.
</p>
<p>
The <code>Gccgo</code> compiler is a front end written in C++
with a recursive descent parser coupled to the
standard GCC back end.
</p>
<p>
Go turned out to be a fine language in which to implement a Go compiler,
although that was not its original goal.
Not being self-hosting from the beginning allowed Go's design to
concentrate on its original use case, which was networked servers.
Had we decided Go should compile itself early on, we might have
ended up with a language targeted more for compiler construction,
which is a worthy goal but not the one we had initially.
</p>
<p>
Although <code>gc</code> does not use them (yet?), a native lexer and
parser are available in the <a href="/pkg/go/"><code>go</code></a> package
and there is also a native <a href="/pkg/go/types">type checker</a>.
</p>
<h3 id="How_is_the_run_time_support_implemented">
How is the run-time support implemented?</h3>
<p>
Again due to bootstrapping issues, the run-time code was originally written mostly in C (with a
tiny bit of assembler) but it has since been translated to Go
(except for some assembler bits).
<code>Gccgo</code>'s run-time support uses <code>glibc</code>.
The <code>gccgo</code> compiler implements goroutines using
a technique called segmented stacks,
supported by recent modifications to the gold linker.
<code>Gollvm</code> similarly is built on the corresponding
LLVM infrastructure.
</p>
<h3 id="Why_is_my_trivial_program_such_a_large_binary">
Why is my trivial program such a large binary?</h3>
<p>
The linker in the <code>gc</code> toolchain
creates statically-linked binaries by default.
All Go binaries therefore include the Go
runtime, along with the run-time type information necessary to support dynamic
type checks, reflection, and even panic-time stack traces.
</p>
<p>
A simple C "hello, world" program compiled and linked statically using
gcc on Linux is around 750 kB, including an implementation of
<code>printf</code>.
An equivalent Go program using
<code>fmt.Printf</code> weighs a couple of megabytes, but that includes
more powerful run-time support and type and debugging information.
</p>
<p>
A Go program compiled with <code>gc</code> can be linked with
the <code>-ldflags=-w</code> flag to disable DWARF generation,
removing debugging information from the binary but with no
other loss of functionality.
This can reduce the binary size substantially.
</p>
<h3 id="unused_variables_and_imports">
Can I stop these complaints about my unused variable/import?</h3>
<p>
The presence of an unused variable may indicate a bug, while
unused imports just slow down compilation,
an effect that can become substantial as a program accumulates
code and programmers over time.
For these reasons, Go refuses to compile programs with unused
variables or imports,
trading short-term convenience for long-term build speed and
program clarity.
</p>
<p>
Still, when developing code, it's common to create these situations
temporarily and it can be annoying to have to edit them out before the
program will compile.
</p>
<p>
Some have asked for a compiler option to turn those checks off
or at least reduce them to warnings.
Such an option has not been added, though,
because compiler options should not affect the semantics of the
language and because the Go compiler does not report warnings, only
errors that prevent compilation.
</p>
<p>
There are two reasons for having no warnings. First, if it's worth
complaining about, it's worth fixing in the code. (And if it's not
worth fixing, it's not worth mentioning.) Second, having the compiler
generate warnings encourages the implementation to warn about weak
cases that can make compilation noisy, masking real errors that
<em>should</em> be fixed.
</p>
<p>
It's easy to address the situation, though. Use the blank identifier
to let unused things persist while you're developing.
</p>
<pre>
import "unused"
// This declaration marks the import as used by referencing an
// item from the package.
var _ = unused.Item // TODO: Delete before committing!
func main() {
debugData := debug.Profile()
_ = debugData // Used only during debugging.
....
}
</pre>
<p>
Nowadays, most Go programmers use a tool,
<a href="https://godoc.org/golang.org/x/tools/cmd/goimports">goimports</a>,
which automatically rewrites a Go source file to have the correct imports,
eliminating the unused imports issue in practice.
This program is easily connected to most editors to run automatically when a Go source file is written.
</p>
<h3 id="virus">
Why does my virus-scanning software think my Go distribution or compiled binary is infected?</h3>
<p>
This is a common occurrence, especially on Windows machines, and is almost always a false positive.
Commercial virus scanning programs are often confused by the structure of Go binaries, which
they don't see as often as those compiled from other languages.
</p>
<p>
If you've just installed the Go distribution and the system reports it is infected, that's certainly a mistake.
To be really thorough, you can verify the download by comparing the checksum with those on the
<a href="https://golang.org/dl/">downloads page</a>.
</p>
<p>
In any case, if you believe the report is in error, please report a bug to the supplier of your virus scanner.
Maybe in time virus scanners can learn to understand Go programs.
</p>
<h2 id="Performance">Performance</h2>
<h3 id="Why_does_Go_perform_badly_on_benchmark_x">
Why does Go perform badly on benchmark X?</h3>
<p>
One of Go's design goals is to approach the performance of C for comparable
programs, yet on some benchmarks it does quite poorly, including several
in <a href="https://go.googlesource.com/exp/+/master/shootout/">golang.org/x/exp/shootout</a>.
The slowest depend on libraries for which versions of comparable performance
are not available in Go.
For instance, <a href="https://go.googlesource.com/exp/+/master/shootout/pidigits.go">pidigits.go</a>
depends on a multi-precision math package, and the C
versions, unlike Go's, use <a href="https://gmplib.org/">GMP</a> (which is
written in optimized assembler).
Benchmarks that depend on regular expressions
(<a href="https://go.googlesource.com/exp/+/master/shootout/regex-dna.go">regex-dna.go</a>,
for instance) are essentially comparing Go's native <a href="/pkg/regexp">regexp package</a> to
mature, highly optimized regular expression libraries like PCRE.
</p>
<p>
Benchmark games are won by extensive tuning and the Go versions of most
of the benchmarks need attention. If you measure comparable C
and Go programs
(<a href="https://go.googlesource.com/exp/+/master/shootout/reverse-complement.go">reverse-complement.go</a>
is one example), you'll see the two languages are much closer in raw performance
than this suite would indicate.
</p>
<p>
Still, there is room for improvement. The compilers are good but could be
better, many libraries need major performance work, and the garbage collector
isn't fast enough yet. (Even if it were, taking care not to generate unnecessary
garbage can have a huge effect.)
</p>
<p>
In any case, Go can often be very competitive.
There has been significant improvement in the performance of many programs
as the language and tools have developed.
See the blog post about
<a href="//blog.golang.org/2011/06/profiling-go-programs.html">profiling
Go programs</a> for an informative example.
<h2 id="change_from_c">Changes from C</h2>
<h3 id="different_syntax">
Why is the syntax so different from C?</h3>
<p>
Other than declaration syntax, the differences are not major and stem
from two desires. First, the syntax should feel light, without too
many mandatory keywords, repetition, or arcana. Second, the language
has been designed to be easy to analyze
and can be parsed without a symbol table. This makes it much easier
to build tools such as debuggers, dependency analyzers, automated
documentation extractors, IDE plug-ins, and so on. C and its
descendants are notoriously difficult in this regard.
</p>
<h3 id="declarations_backwards">
Why are declarations backwards?</h3>
<p>
They're only backwards if you're used to C. In C, the notion is that a
variable is declared like an expression denoting its type, which is a
nice idea, but the type and expression grammars don't mix very well and
the results can be confusing; consider function pointers. Go mostly
separates expression and type syntax and that simplifies things (using
prefix <code>*</code> for pointers is an exception that proves the rule). In C,
the declaration
</p>
<pre>
int* a, b;
</pre>
<p>
declares <code>a</code> to be a pointer but not <code>b</code>; in Go
</p>
<pre>
var a, b *int
</pre>
<p>
declares both to be pointers. This is clearer and more regular.
Also, the <code>:=</code> short declaration form argues that a full variable
declaration should present the same order as <code>:=</code> so
</p>
<pre>
var a uint64 = 1
</pre>
<p>
has the same effect as
</p>
<pre>
a := uint64(1)
</pre>
<p>
Parsing is also simplified by having a distinct grammar for types that
is not just the expression grammar; keywords such as <code>func</code>
and <code>chan</code> keep things clear.
</p>
<p>
See the article about
<a href="/doc/articles/gos_declaration_syntax.html">Go's Declaration Syntax</a>
for more details.
</p>
<h3 id="no_pointer_arithmetic">
Why is there no pointer arithmetic?</h3>
<p>
Safety. Without pointer arithmetic it's possible to create a
language that can never derive an illegal address that succeeds
incorrectly. Compiler and hardware technology have advanced to the
point where a loop using array indices can be as efficient as a loop
using pointer arithmetic. Also, the lack of pointer arithmetic can
simplify the implementation of the garbage collector.
</p>
<h3 id="inc_dec">
Why are <code>++</code> and <code>--</code> statements and not expressions? And why postfix, not prefix?</h3>
<p>
Without pointer arithmetic, the convenience value of pre- and postfix
increment operators drops. By removing them from the expression
hierarchy altogether, expression syntax is simplified and the messy
issues around order of evaluation of <code>++</code> and <code>--</code>
(consider <code>f(i++)</code> and <code>p[i] = q[++i]</code>)
are eliminated as well. The simplification is
significant. As for postfix vs. prefix, either would work fine but
the postfix version is more traditional; insistence on prefix arose
with the STL, a library for a language whose name contains, ironically, a
postfix increment.
</p>
<h3 id="semicolons">
Why are there braces but no semicolons? And why can't I put the opening
brace on the next line?</h3>
<p>
Go uses brace brackets for statement grouping, a syntax familiar to
programmers who have worked with any language in the C family.
Semicolons, however, are for parsers, not for people, and we wanted to
eliminate them as much as possible. To achieve this goal, Go borrows
a trick from BCPL: the semicolons that separate statements are in the
formal grammar but are injected automatically, without lookahead, by
the lexer at the end of any line that could be the end of a statement.
This works very well in practice but has the effect that it forces a
brace style. For instance, the opening brace of a function cannot
appear on a line by itself.
</p>
<p>
Some have argued that the lexer should do lookahead to permit the
brace to live on the next line. We disagree. Since Go code is meant
to be formatted automatically by
<a href="/cmd/gofmt/"><code>gofmt</code></a>,
<i>some</i> style must be chosen. That style may differ from what
you've used in C or Java, but Go is a different language and
<code>gofmt</code>'s style is as good as any other. More
important&mdash;much more important&mdash;the advantages of a single,
programmatically mandated format for all Go programs greatly outweigh
any perceived disadvantages of the particular style.
Note too that Go's style means that an interactive implementation of
Go can use the standard syntax one line at a time without special rules.
</p>
<h3 id="garbage_collection">
Why do garbage collection? Won't it be too expensive?</h3>
<p>
One of the biggest sources of bookkeeping in systems programs is
managing the lifetimes of allocated objects.
In languages such as C in which it is done manually,
it can consume a significant amount of programmer time and is
often the cause of pernicious bugs.
Even in languages like C++ or Rust that provide mechanisms
to assist, those mechanisms can have a significant effect on the
design of the software, often adding programming overhead
of its own.
We felt it was critical to eliminate such
programmer overheads, and advances in garbage collection
technology in the last few years gave us confidence that it
could be implemented cheaply enough, and with low enough
latency, that it could be a viable approach for networked
systems.
</p>
<p>
Much of the difficulty of concurrent programming
has its roots in the object lifetime problem:
as objects get passed among threads it becomes cumbersome
to guarantee they become freed safely.
Automatic garbage collection makes concurrent code far easier to write.
Of course, implementing garbage collection in a concurrent environment is
itself a challenge, but meeting it once rather than in every
program helps everyone.
</p>
<p>
Finally, concurrency aside, garbage collection makes interfaces
simpler because they don't need to specify how memory is managed across them.
</p>
<p>
This is not to say that the recent work in languages
like Rust that bring new ideas to the problem of managing
resources is misguided; we encourage this work and are excited to see
how it evolves.
But Go takes a more traditional approach by addressing
object lifetimes through
garbage collection, and garbage collection alone.
</p>
<p>
The current implementation is a mark-and-sweep collector.
If the machine is a multiprocessor, the collector runs on a separate CPU
core in parallel with the main program.
Major work on the collector in recent years has reduced pause times
often to the sub-millisecond range, even for large heaps,
all but eliminating one of the major objections to garbage collection
in networked servers.
Work continues to refine the algorithm, reduce overhead and
latency further, and to explore new approaches.
The 2018
<a href="https://talks.golang.org/2018/ismmkeynote">ISMM keynote</a>
by Rick Hudson of the Go team
describes the progress so far and suggests some future approaches.
</p>
<p>
On the topic of performance, keep in mind that Go gives the programmer
considerable control over memory layout and allocation, much more than
is typical in garbage-collected languages. A careful programmer can reduce
the garbage collection overhead dramatically by using the language well;
see the article about
<a href="//blog.golang.org/2011/06/profiling-go-programs.html">profiling
Go programs</a> for a worked example, including a demonstration of Go's
profiling tools.
</p>