1
0
mirror of https://github.com/golang/go synced 2024-11-07 09:06:20 -07:00
go/doc/go1.18.html

1228 lines
52 KiB
HTML
Raw Normal View History

<!--{
"Title": "Go 1.18 Release Notes",
"Path": "/doc/go1.18"
}-->
<!--
NOTE: In this document and others in this directory, the convention is to
set fixed-width phrases with non-fixed-width spaces, as in
<code>hello</code> <code>world</code>.
Do not send CLs removing the interior tags from such phrases.
-->
<style>
main ul li { margin: 0.5em 0; }
</style>
<h2 id="introduction">DRAFT RELEASE NOTES — Introduction to Go 1.18</h2>
<p>
<strong>
Go 1.18 is not yet released. These are work-in-progress
release notes. Go 1.18 is expected to be released in February 2022.
</strong>
</p>
<h2 id="language">Changes to the language</h2>
<h3 id="generics">Generics</h3>
<p><!-- https://golang.org/issue/43651, https://golang.org/issue/45346 -->
Go 1.18 includes an implementation of generic features as described by the
<a href="https://go.googlesource.com/proposal/+/refs/heads/master/design/43651-type-parameters.md">Type
Parameters Proposal</a>.
This includes major - but fully backward-compatible - changes to the language.
</p>
<p>
These new language changes required a large amount of new code that
has not had significant testing in production settings. That will
only happen as more people write and use generic code. We believe
that this feature is well implemented and high quality. However,
unlike most aspects of Go, we can't back up that belief with real
world experience. Therefore, while we encourage the use of generics
where it makes sense, please use appropriate caution when deploying
generic code in production.
</p>
<p>
The following is a list of the most visible changes. For a more comprehensive overview, see the
<a href="https://go.googlesource.com/proposal/+/refs/heads/master/design/43651-type-parameters.md">proposal</a>.
For details see the <a href="/ref/spec">language spec</a>.
</p>
<ul>
<li>
The syntax for
<a href="/ref/spec#Function_declarations">function</a> and
<a href="/ref/spec#Type_declarations">type declarations</a>
now accepts
<a href="/ref/spec#Type_parameters">type parameters</a>.
</li>
<li>
Parameterized functions and types can be instantiated by following them with a list of
type arguments in square brackets.
</li>
<li>
The new token <code>~</code> has been added to the set of
<a href="/ref/spec#Operators_and_punctuation">operators and punctuation</a>.
</li>
<li>
The syntax for
<a href="/ref/spec#Interface_types">Interface types</a>
now permits the embedding of arbitrary types (not just type names of interfaces)
as well as union and <code>~T</code> type elements. Such interfaces may only be used
as type constraints.
An interface now defines a set of types as well as a set of methods.
</li>
<li>
The new
<a href="/ref/spec#Predeclared_identifiers">predeclared identifier</a>
<code>any</code> is an alias for the empty interface. It may be used instead of
<code>interface{}</code>.
</li>
<li>
The new
<a href="/ref/spec#Predeclared_identifiers">predeclared identifier</a>
<code>comparable</code> is an interface the denotes the set of all types which can be
compared using <code>==</code> or <code>!=</code>. It may only be used as (or embedded in)
a type constraint.
</li>
</ul>
<p>
There are three experimental packages using generics that may be
useful.
These packages are in x/exp repository; their API is not covered by
the Go 1 guarantee and may change as we gain more experience with
generics.
<dl>
<dt><a href="https://pkg.go.dev/golang.org/x/exp/constraints"><code>golang.org/x/exp/constraints</code></a></dt>
<dd>
<p>
Constraints that are useful for generic code, such as
<a href="https://pkg.go.dev/golang.org/x/exp/constraints#Ordered"><code>constraints.Ordered</code></a>.
</p>
</dd>
<dt><a href="https://pkg.go.dev/golang.org/x/exp/slices"><code>golang.org/x/exp/slices</code></a></dt>
<dd>
<p>
A collection of generic functions that operate on slices of
any element type.
</p>
</dd>
<dt><a href="https://pkg.go.dev/golang.org/x/exp/maps"><code>golang.org/x/exp/maps</code></a></dt>
<dd>
<p>
A collection of generic functions that operate on maps of
any key or element type.
</p>
</dd>
</dl>
</p>
<p>
The current generics implementation has the following limitations:
<ul>
<li><!-- https://golang.org/issue/47631 -->
The Go compiler cannot currently handle type declarations inside generic functions
or methods. We hope to provide support for this feature in Go 1.19.
</li>
<li><!-- https://golang.org/issue/50937 -->
The Go compiler currently does not accept arguments of type parameter type with
the predeclared functions <code>real</code>, <code>imag</code>, and <code>complex</code>.
We hope to remove this restriction in Go 1.19.
</li>
<li><!-- https://golang.org/issue/49030 -->
Embedding a type parameter, or a pointer to a type parameter, as
an unnamed field in a struct type is not permitted. Similarly
embedding a type parameter in an interface type is not permitted.
Whether these will ever be permitted is unclear at present.
</li>
<li>
A union element with more than one term may not contain an
interface type with a non-empty method set. Whether this will
ever be permitted is unclear at present.
</li>
</ul>
</p>
<h3 id="bug_fixes">Bug fixes</h3>
<p>
The Go 1.18 compiler now correctly reports <code>declared but not used</code> errors
for variables that are set inside a function literal but are never used. Before Go 1.18,
the compiler did not report an error in such cases. This fixes long-outstanding compiler
issue <a href="https://golang.org/issue/8560">#8560</a>. As a result of this change,
(possibly incorrect) programs may not compile anymore. The necessary fix is
straightforward: fix the program if it was in fact incorrect, or use the offending
variable, for instance by assigning it to the blank identifier <code>_</code>.
Since <code>go vet</code> always pointed out this error, the number of affected
programs is likely very small.
</p>
<p>
The Go 1.18 compiler now reports an overflow when passing a rune constant expression
such as <code>'1' &lt;&lt; 32</code> as an argument to the predeclared functions
<code>print</code> and <code>println</code>, consistent with the behavior of
user-defined functions. Before Go 1.18, the compiler did not report an error
in such cases but silently accepted such constant arguments if they fit into an
<code>int64</code>. As a result of this change, (possibly incorrect) programs
may not compile anymore. The necessary fix is straightforward: fix the program if it
was in fact incorrect, or explicitly convert the offending argument to the correct type.
Since <code>go vet</code> always pointed out this error, the number of affected
programs is likely very small.
</p>
<h2 id="ports">Ports</h2>
<h3 id="amd64">AMD64</h3>
<p><!-- CL 349595 -->
Go 1.18 introduces the new <code>GOAMD64</code> environment variable, which selects at compile time
a mininum target version of the AMD64 architecture. Allowed values are <code>v1</code>,
<code>v2</code>, <code>v3</code>, or <code>v4</code>. Each higher level requires,
and takes advantage of, additional processor features. A detailed
description can be found
<a href="https://golang.org/wiki/MinimumRequirements#amd64">here</a>.
</p>
<p>
The <code>GOAMD64</code> environment variable defaults to <code>v1</code>.
</p>
<h3 id="riscv">RISC-V</h3>
<p><!-- golang.org/issue/47100, CL 334872 -->
The 64-bit RISC-V architecture on Linux (the <code>linux/riscv64</code> port)
now supports the <code>c-archive</code> and <code>c-shared</code> build modes.
</p>
<h3 id="windows">Windows</h3>
<p><!-- https://golang.org/issue/49759 -->
The <code>windows/arm</code> and <code>windows/arm64</code> ports now support
non-cooperative preemption, bringing that capability to all four Windows
ports, which should hopefully address subtle bugs encountered when calling
into Win32 functions that block for extended periods of time.
</p>
<h3 id="ios">iOS</h3>
<p><!-- golang.org/issue/48076, golang.org/issue/49616 -->
On iOS (the <code>ios/arm64</code> port)
and iOS simulator running on AMD64-based macOS (the <code>ios/amd64</code> port),
Go 1.18 now requires iOS 12 or later; support for previous versions has been discontinued.
</p>
<h3 id="freebsd">FreeBSD</h3>
<p>
Go 1.18 is the last release that is supported on FreeBSD 11.x, which has
already reached end-of-life. Go 1.19 will require FreeBSD 12.2+ or FreeBSD
13.0+.
FreeBSD 13.0+ will require a kernel with the COMPAT_FREEBSD12 option set (this is the default).
</p>
<h2 id="tools">Tools</h2>
<h3 id="fuzzing">Fuzzing</h3>
<p>
Go 1.18 includes an implementation of fuzzing as described by
<a href="https://golang.org/issue/44551">the fuzzing proposal</a>.
</p>
<p>
See the <a href="https://go.dev/doc/fuzz">fuzzing landing page</a> to get
started.
</p>
<p>
Please be aware that fuzzing can consume a lot of memory and may impact your
machines performance while it runs. Also be aware that the fuzzing engine
writes values that expand test coverage to a fuzz cache directory within
<code>$GOCACHE/fuzz</code> while it runs. There is currently no limit to the
number of files or total bytes that may be written to the fuzz cache, so it
may occupy a large amount of storage (possibly several GBs).
</p>
<h3 id="go-command">Go command</h3>
<p><!-- golang.org/issue/43684 -->
<code>go</code> <code>get</code> no longer builds or installs packages in
module-aware mode. <code>go</code> <code>get</code> is now dedicated to
adjusting dependencies in <code>go.mod</code>. Effectively, the
<code>-d</code> flag is always enabled. To install the latest version
of an executable outside the context of the current module, use
<a href="https://golang.org/ref/mod#go-install"><code>go</code>
<code>install</code> <code>example.com/cmd@latest</code></a>. Any
<a href="https://golang.org/ref/mod#version-queries">version query</a>
may be used instead of <code>latest</code>. This form of <code>go</code>
<code>install</code> was added in Go 1.16, so projects supporting older
versions may need to provide install instructions for both <code>go</code>
<code>install</code> and <code>go</code> <code>get</code>. <code>go</code>
<code>get</code> now reports an error when used outside a module, since there
is no <code>go.mod</code> file to update. In GOPATH mode (with
<code>GO111MODULE=off</code>), <code>go</code> <code>get</code> still builds
and installs packages, as before.
</p>
<p><!-- golang.org/issue/37475 -->
The <code>go</code> command now embeds version control information in
binaries including the currently checked-out revision, commit time, and a
flag indicating whether edited or untracked files are present. Version
control information is embedded if the <code>go</code> command is invoked in
a directory within a Git, Mercurial, Fossil, or Bazaar repository, and the
<code>main</code> package and its containing main module are in the same
repository. This information may be omitted using the flag
<code>-buildvcs=false</code>.
</p>
<p><!-- golang.org/issue/37475 -->
Additionally, the <code>go</code> command embeds information about the build
including build and tool tags (set with <code>-tags</code>), compiler,
assembler, and linker flags (like <code>-gcflags</code>), whether cgo was
enabled, and if it was, the values of the cgo environment variables
(like <code>CGO_CFLAGS</code>).
Both VCS and build information may be read together with module
information using
<code>go</code> <code>version</code> <code>-m</code> <code>file</code> or
<code>runtime/debug.ReadBuildInfo</code> (for the currently running binary)
or the new <a href="#debug/buildinfo"><code>debug/buildinfo</code></a>
package.
</p>
<p><!-- CL 369977 -->
The underlying data format of the embedded build information can change with
new go releases, so an older version of <code>go</code> may not handle the
build information produced with a newer version of <code>go</code>.
To read the version information from a binary built with <code>go</code> 1.18,
use the <code>go</code> <code>version</code> command and the
<code>debug/buildinfo</code> package from <code>go</code> 1.18+.
</p>
<p><!-- https://golang.org/issue/44435 -->
If the main module's <code>go.mod</code> file
specifies <a href="/ref/mod#go-mod-file-go"><code>go</code> <code>1.17</code></a>
or higher, <code>go</code> <code>mod</code> <code>download</code> without
arguments now downloads source code for only the modules
explicitly <a href="/ref/mod#go-mod-file-require">required</a> in the main
module's <code>go.mod</code> file. (In a <code>go</code> <code>1.17</code> or
higher module, that set already includes all dependencies needed to build the
packages and tests in the main module.)
To also download source code for transitive dependencies, use
<code>go</code> <code>mod</code> <code>download</code> <code>all</code>.
</p>
<p><!-- https://golang.org/issue/47327 -->
The <code>go</code> <code>mod</code> <code>vendor</code> subcommand now
supports a <code>-o</code> flag to set the output directory.
(Other <code>go</code> commands still read from the <code>vendor</code>
directory at the module root when loading packages
with <code>-mod=vendor</code>, so the main use for this flag is for
third-party tools that need to collect package source code.)
</p>
<p><!-- CL 298612 -->
The <code>go</code> <code>build</code> command and related commands
now support an <code>-asan</code> flag that enables interoperation
with C (or C++) code compiled with the address sanitizer (C compiler
option <code>-fsanitize=address</code>).
</p>
<p><!-- https://golang.org/issue/47738 -->
The <code>go</code> <code>mod</code> <code>tidy</code> command now retains
additional checksums in the <code>go.sum</code> file for modules whose source
code is needed to verify that each imported package is provided by only one
module in the <a href="/ref/mod#glos-build-list">build list</a>. Because this
condition is rare and failure to apply it results in a build error, this
change is <em>not</em> conditioned on the <code>go</code> version in the main
module's <code>go.mod</code> file.
</p>
<p><!-- https://golang.org/issue/45713 -->
The <code>go</code> command now supports a "Workspace" mode. If a
<code>go.work</code> file is found in the working directory or a
parent directory, or one is specified using the <code>-workfile</code>
flag, it will put the <code>go</code> command into workspace mode.
In workspace mode, the <code>go.work</code> file will be used to
determine the set of main modules used as the roots for module
resolution, instead of using the normally-found <code>go.mod</code>
file to specify the single main module. For more information see the
<a href="/pkg/cmd/go#hdr-Workspace_maintenance"><code>go work</code></a>
documentation.
</p>
<p><!-- CL 251441 -->
The <code>go</code> command now supports additional command line
options for the new <a href="#fuzzing">fuzzing support described
above</a>:
<ul>
<li>
<code>go test</code> supports
<code>-fuzz</code>, <code>-fuzztime</code>, and
<code>-fuzzminimizetime</code> options.
For documentation on these see
<a href="/pkg/cmd/go#hdr-Testing_flags"><code>go help testflag</code></a>.
</li>
<li>
<code>go clean</code> supports a <code>-fuzzcache</code>
option.
For documentation see
<a href="/pkg/cmd/go#hdr-Remove_object_files_and_cached_files"><code>go help clean</code></a>.
</li>
</ul>
</p>
cmd/gofmt: format files in parallel gofmt is pretty heavily CPU-bound, since parsing and formatting 1MiB of Go code takes much longer than reading that amount of bytes from disk. However, parsing and manipulating a large Go source file is very difficult to parallelize, so we continue to process each file in its own goroutine. A Go module may contain a large number of Go source files, so we need to bound the amount of work in flight. However, because the distribution of sizes for Go source files varies widely — from tiny doc.go files containing a single package comment all the way up to massive API wrappers generated by automated tools — the amount of time, work, and memory overhead needed to process each file also varies. To account for this variability, we limit the in-flight work by bytes of input rather than by number of files. That allows us to make progress on many small files while we wait for work on a handful of large files to complete. The gofmt tool has a well-defined output format on stdout, which was previously deterministic. We keep it deterministic by printing the results of each file in order, using a lazily-synchronized io.Writer (loosly inspired by Haskell's IO monad). After a file has been formatted in memory, we keep it in memory (again, limited by the corresponding number of input bytes) until the output for all previous files has been flushed. This adds a bit of latency compared to emitting the output in nondeterministic order, but a little extra latency seems worth the cost to preserve output stability. This change is based on Daniel Martí's work in CL 284139, but using a weighted semaphore and ephemeral goroutines instead of a worker pool and batches. Benchmark results are similar, and I find the concurrency in this approach a bit easier to reason about. In the batching-based approach, the batch size allows us to "look ahead" to find large files and start processing them early. To keep the CPUs saturated and prevent stragglers, we would need to tune the batch size to be about the same as the largest input files. If the batch size is set too high, a large batch of small files could turn into a straggler, but if the batch size is set too low, the largest files in the data set won't be started early enough and we'll end up with a large-file straggler. One possible alternative would be to sort by file size instead of batching: identify all of the files to be processed, sort from largest to smallest, and then process the largest files first so that the "tail" of processing covers the smallest files. However, that approach would still fail to saturate available CPU when disk latency is high, would require buffering an arbitrary amount of metadata in order to sort by size, and (perhaps most importantly!) would not allow the `gofmt` binary to preserve the same (deterministic) output order that it has today. In contrast, with a semaphore we can produce the same deterministic output as ever using only one tuning parameter: the memory footprint, expressed as a rough lower bound on the amount of RAM available per thread. While we're below the memory limit, we can run arbitrarily many disk operations arbitrarily far ahead, and process the results of those operations whenever they become avaliable. Then it's up to the kernel (not us) to schedule the disk operations for throughput and latency, and it's up to the runtime (not us) to schedule the goroutines so that they complete quickly. In practice, even a modest assumption of a few megabytes per thread seems to provide a nice speedup, and it should scale reasonably even to machines with vastly different ratios of CPU to disk. (In practice, I expect that most 'gofmt' invocations will work with files on at most one physical disk, so the CPU:disk ratio should vary more-or-less directly with the thread count, whereas the CPU:memory ratio is more-or-less independent of thread count.) name \ time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 11.9s ± 2% 2.7s ± 3% 2.8s ± 5% name \ user-time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 13.5s ± 2% 14.4s ± 1% 14.7s ± 1% name \ sys-time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 465ms ± 8% 229ms ±28% 232ms ±31% name \ peak-RSS-bytes baseline.txt 284139.txt simplified.txt GofmtGorootCmd 77.7MB ± 4% 162.2MB ±10% 192.9MB ±15% For #43566 Change-Id: I4ba251eb4d188a3bd1901039086be57f0b341910 Reviewed-on: https://go-review.googlesource.com/c/go/+/317975 Trust: Bryan C. Mills <bcmills@google.com> Trust: Daniel Martí <mvdan@mvdan.cc> Run-TryBot: Bryan C. Mills <bcmills@google.com> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
2021-01-16 11:03:31 -07:00
<h3 id="gofmt"><code>gofmt</code></h3>
<p><!-- https://golang.org/issue/43566 -->
<code>gofmt</code> now reads and formats input files concurrently, with a
memory limit proportional to <code>GOMAXPROCS</code>. On a machine with
multiple CPUs, <code>gofmt</code> should now be significantly faster.
cmd/gofmt: format files in parallel gofmt is pretty heavily CPU-bound, since parsing and formatting 1MiB of Go code takes much longer than reading that amount of bytes from disk. However, parsing and manipulating a large Go source file is very difficult to parallelize, so we continue to process each file in its own goroutine. A Go module may contain a large number of Go source files, so we need to bound the amount of work in flight. However, because the distribution of sizes for Go source files varies widely — from tiny doc.go files containing a single package comment all the way up to massive API wrappers generated by automated tools — the amount of time, work, and memory overhead needed to process each file also varies. To account for this variability, we limit the in-flight work by bytes of input rather than by number of files. That allows us to make progress on many small files while we wait for work on a handful of large files to complete. The gofmt tool has a well-defined output format on stdout, which was previously deterministic. We keep it deterministic by printing the results of each file in order, using a lazily-synchronized io.Writer (loosly inspired by Haskell's IO monad). After a file has been formatted in memory, we keep it in memory (again, limited by the corresponding number of input bytes) until the output for all previous files has been flushed. This adds a bit of latency compared to emitting the output in nondeterministic order, but a little extra latency seems worth the cost to preserve output stability. This change is based on Daniel Martí's work in CL 284139, but using a weighted semaphore and ephemeral goroutines instead of a worker pool and batches. Benchmark results are similar, and I find the concurrency in this approach a bit easier to reason about. In the batching-based approach, the batch size allows us to "look ahead" to find large files and start processing them early. To keep the CPUs saturated and prevent stragglers, we would need to tune the batch size to be about the same as the largest input files. If the batch size is set too high, a large batch of small files could turn into a straggler, but if the batch size is set too low, the largest files in the data set won't be started early enough and we'll end up with a large-file straggler. One possible alternative would be to sort by file size instead of batching: identify all of the files to be processed, sort from largest to smallest, and then process the largest files first so that the "tail" of processing covers the smallest files. However, that approach would still fail to saturate available CPU when disk latency is high, would require buffering an arbitrary amount of metadata in order to sort by size, and (perhaps most importantly!) would not allow the `gofmt` binary to preserve the same (deterministic) output order that it has today. In contrast, with a semaphore we can produce the same deterministic output as ever using only one tuning parameter: the memory footprint, expressed as a rough lower bound on the amount of RAM available per thread. While we're below the memory limit, we can run arbitrarily many disk operations arbitrarily far ahead, and process the results of those operations whenever they become avaliable. Then it's up to the kernel (not us) to schedule the disk operations for throughput and latency, and it's up to the runtime (not us) to schedule the goroutines so that they complete quickly. In practice, even a modest assumption of a few megabytes per thread seems to provide a nice speedup, and it should scale reasonably even to machines with vastly different ratios of CPU to disk. (In practice, I expect that most 'gofmt' invocations will work with files on at most one physical disk, so the CPU:disk ratio should vary more-or-less directly with the thread count, whereas the CPU:memory ratio is more-or-less independent of thread count.) name \ time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 11.9s ± 2% 2.7s ± 3% 2.8s ± 5% name \ user-time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 13.5s ± 2% 14.4s ± 1% 14.7s ± 1% name \ sys-time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 465ms ± 8% 229ms ±28% 232ms ±31% name \ peak-RSS-bytes baseline.txt 284139.txt simplified.txt GofmtGorootCmd 77.7MB ± 4% 162.2MB ±10% 192.9MB ±15% For #43566 Change-Id: I4ba251eb4d188a3bd1901039086be57f0b341910 Reviewed-on: https://go-review.googlesource.com/c/go/+/317975 Trust: Bryan C. Mills <bcmills@google.com> Trust: Daniel Martí <mvdan@mvdan.cc> Run-TryBot: Bryan C. Mills <bcmills@google.com> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
2021-01-16 11:03:31 -07:00
</p>
<h3 id="vet"><code>vet</code></h3>
<h4 id="vet-generics">Updates for Generics</h4>
<p><!-- https://golang.org/issue/48704 -->
The <code>vet</code> tool is updated to support generic code. In most cases,
it reports an error in generic code whenever it would report an error in the
equivalent non-generic code after substituting for type parameters with a
type from their
<a href="https://golang.org/ref/spec#Interface_types">type set</a>.
For example, <code>vet</code> reports a format error in
<pre>func Print[T ~int|~string](t T) {
fmt.Printf("%d", t)
}</pre>
because it would report a format error in the non-generic equivalent of
<code>Print[string]</code>:
<pre>func PrintString(x string) {
fmt.Printf("%d", x)
}</pre>
</p>
<h4 id="vet-precision">Precision improvements for existing checkers</h4>
<p><!-- CL 323589 356830 319689 355730 351553 338529 -->
The <code>cmd/vet</code> checkers <code>copylock</code>, <code>printf</code>,
<code>sortslice</code>, <code>testinggoroutine</code>, and <code>tests</code>
have all had moderate precision improvements to handle additional code patterns.
This may lead to newly reported errors in existing packages. For example, the
<code>printf</code> checker now tracks formatting strings created by
concatenating string constants. So <code>vet</code> will report an error in:
<pre>
// fmt.Printf formatting directive %d is being passed to Println.
fmt.Println("%d"+` ≡ x (mod 2)`+"\n", x%2)
</pre>
</p>
<h2 id="runtime">Runtime</h2>
<p><!-- https://golang.org/issue/44167 -->
The garbage collector now includes non-heap sources of garbage collector work
(e.g., stack scanning) when determining how frequently to run. As a result,
garbage collector overhead is more predictable when these sources are
significant. For most applications these changes will be negligible; however,
some Go applications may now use less memory and spend more time on garbage
collection, or vice versa, than before. The intended workaround is to tweak
<code>GOGC</code> where necessary.
</p>
<p><!-- CL 358675, CL 353975, CL 353974 -->
The runtime now returns memory to the operating system more efficiently and has
been tuned to work more aggressively as a result.
</p>
<p><!-- CL 352057, https://golang.org/issue/45728 -->
Go 1.17 generally improved the formatting of arguments in stack traces,
but could print inaccurate values for arguments passed in registers.
This is improved in Go 1.18 by printing a question mark (<code>?</code>)
after each value that may be inaccurate.
</p>
<h2 id="compiler">Compiler</h2>
<p><!-- https://golang.org/issue/40724 -->
Go 1.17 <a href="go1.17#compiler">implemented</a> a new way of passing
function arguments and results using registers instead of the stack
on 64-bit x86 architecture on selected operating systems.
Go 1.18 expands the supported platforms to include 64-bit ARM (<code>GOARCH=arm64</code>),
big- and little-endian 64-bit PowerPC (<code>GOARCH=ppc64</code>, <code>ppc64le</code>),
as well as 64-bit x86 architecture (<code>GOARCH=amd64</code>)
on all operating systems.
On 64-bit ARM and 64-bit PowerPC systems, benchmarking shows
typical performance improvements of 10% or more.
</p>
<p>
As <a href="go1.17#compiler">mentioned</a> in the Go 1.17 release notes,
this change does not affect the functionality of any safe Go code and
is designed to have no impact on most assembly code. See the
<a href="go1.17#compiler">Go 1.17 release notes</a> for more details.
</p>
<p><!-- CL 355497, CL 356869 -->
The compiler now can inline functions that contain range loops or
labeled for loops.
</p>
<p><!-- CL 298611 -->
The new compiler <code>-asan</code> option supports the
new <code>go</code> command <code>-asan</code> option.
</p>
<p><!-- https://golang.org/issue/50954 -->
Because the compiler's type checker was replaced in its entirety to
support generics, some error messages now may use different wording
than before. In some cases, pre-Go 1.18 error messages provided more
detail or were phrased in a more helpful way.
We intend to address these cases in Go 1.19.
</p>
<p> <!-- https://github.com/golang/go/issues/49569 -->
Because of changes in the compiler related to supporting generics, the
Go 1.18 compile speed can be roughly 15% slower than the Go 1.17 compile speed.
The execution time of the compiled code is not affected. We
intend to improve the speed of the compiler in Go 1.19.
</p>
<h2 id="linker">Linker</h2>
<p><!-- CL 298610 -->
The new linker <code>-asan</code> option supports the
new <code>go</code> command <code>-asan</code> option.
</p>
<h2 id="library">Core library</h2>
<h3 id="debug/buildinfo">New <code>debug/buildinfo</code> package</h3>
<p><!-- golang.org/issue/39301 -->
The new <a href="/pkg/debug/buildinfo"><code>debug/buildinfo</code></a> package
provides access to module versions, version control information, and build
flags embedded in executable files built by the <code>go</code> command.
The same information is also available via
<a href="/pkg/runtime/debug#ReadBuildInfo"><code>runtime/debug.ReadBuildInfo</code></a>
for the currently running binary and via <code>go</code>
<code>version</code> <code>-m</code> on the command line.
</p>
<h3 id="netip">New <code>net/netip</code> package</h3>
<p>
The new <a href="/pkg/net/netip/"><code>net/netip</code></a>
package defines a new IP address type, <a href="/pkg/net/netip/#Addr"><code>Addr</code></a>.
Compared to the existing
<a href="/pkg/net/#IP"><code>net.IP</code></a> type, the <code>netip.Addr</code> type takes less
memory, is immutable, and is comparable so it supports <code>==</code>
and can be used as a map key.
</p>
<p>
In addition to <code>Addr</code>, the package defines
<a href="/pkg/net/netip/#AddrPort"><code>AddrPort</code></a>, representing
an IP and port, and
<a href="/pkg/net/netip/#Prefix"><code>Prefix</code></a>, representing
a network CIDR prefix.
</p>
<p>
The package also defines several functions to create and examine
these new types:
<a href="/pkg/net/netip#AddrFrom4"><code>AddrFrom4</code></a>,
<a href="/pkg/net/netip#AddrFrom16"><code>AddrFrom16</code></a>,
<a href="/pkg/net/netip#AddrFromSlice"><code>AddrFromSlice</code></a>,
<a href="/pkg/net/netip#AddrPortFrom"><code>AddrPortFrom</code></a>,
<a href="/pkg/net/netip#IPv4Unspecified"><code>IPv4Unspecified</code></a>,
<a href="/pkg/net/netip#IPv6LinkLocalAllNodes"><code>IPv6LinkLocalAllNodes</code></a>,
<a href="/pkg/net/netip#IPv6Unspecified"><code>IPv6Unspecified</code></a>,
<a href="/pkg/net/netip#MustParseAddr"><code>MustParseAddr</code></a>,
<a href="/pkg/net/netip#MustParseAddrPort"><code>MustParseAddrPort</code></a>,
<a href="/pkg/net/netip#MustParsePrefix"><code>MustParsePrefix</code></a>,
<a href="/pkg/net/netip#ParseAddr"><code>ParseAddr</code></a>,
<a href="/pkg/net/netip#ParseAddrPort"><code>ParseAddrPort</code></a>,
<a href="/pkg/net/netip#ParsePrefix"><code>ParsePrefix</code></a>,
<a href="/pkg/net/netip#PrefixFrom"><code>PrefixFrom</code></a>.
</p>
<p>
The <a href="/pkg/net/"><code>net</code></a> package includes new
methods that parallel existing methods, but
return <code>netip.AddrPort</code> instead of the
heavier-weight <a href="/pkg/net/#IP"><code>net.IP</code></a> or
<a href="/pkg/net/#UDPAddr"><code>*net.UDPAddr</code></a> types:
<a href="/pkg/net/#Resolver.LookupNetIP"><code>Resolver.LookupNetIP</code></a>,
<a href="/pkg/net/#UDPConn.ReadFromUDPAddrPort"><code>UDPConn.ReadFromUDPAddrPort</code></a>,
<a href="/pkg/net/#UDPConn.ReadMsgUDPAddrPort"><code>UDPConn.ReadMsgUDPAddrPort</code></a>,
<a href="/pkg/net/#UDPConn.WriteToUDPAddrPort"><code>UDPConn.WriteToUDPAddrPort</code></a>,
<a href="/pkg/net/#UDPConn.WriteMsgUDPAddrPort"><code>UDPConn.WriteMsgUDPAddrPort</code></a>.
The new <code>UDPConn</code> methods support allocation-free I/O.
</p>
<p>
The <code>net</code> package also now includes functions and methods
to convert between the existing
<a href="/pkg/net/#TCPAddr"><code>TCPAddr</code></a>/<a href="/pkg/net/#UDPAddr"><code>UDPAddr</code></a>
types and <code>netip.AddrPort</code>:
<a href="/pkg/net/#TCPAddrFromAddrPort"><code>TCPAddrFromAddrPort</code></a>,
<a href="/pkg/net/#UDPAddrFromAddrPort"><code>UDPAddrFromAddrPort</code></a>,
<a href="/pkg/net/#TCPAddr.AddrPort"><code>TCPAddr.AddrPort</code></a>,
<a href="/pkg/net/#UDPAddr.AddrPort"><code>UDPAddr.AddrPort</code></a>.
</p>
<h3 id="tls10">TLS 1.0 and 1.1 disabled by default client-side</h3>
<p><!-- CL 359779, golang.org/issue/45428 -->
If <a href="/pkg/crypto/tls/#Config.MinVersion"><code>Config.MinVersion</code></a>
is not set, it now defaults to TLS 1.2 for client connections. Any safely
up-to-date server is expected to support TLS 1.2, and browsers have required
it since 2020. TLS 1.0 and 1.1 are still supported by setting
<code>Config.MinVersion</code> to <code>VersionTLS10</code>.
The server-side default is unchanged at TLS 1.0.
</p>
<p>
The default can be temporarily reverted to TLS 1.0 by setting the
<code>GODEBUG=tls10default=1</code> environment variable.
This option will be removed in Go 1.19.
</p>
<h3 id="sha1">Rejecting SHA-1 certificates</h3>
<p><!-- CL 359777, golang.org/issue/41682 -->
<code>crypto/x509</code> will now
reject certificates signed with the SHA-1 hash function. This doesn't
apply to self-signed root certificates. Practical attacks against SHA-1
<a href="https://shattered.io/">have been demonstrated since 2017</a> and publicly
trusted Certificate Authorities have not issued SHA-1 certificates since 2015.
</p>
<p>
This can be temporarily reverted by setting the
<code>GODEBUG=x509sha1=1</code> environment variable.
This option will be removed in Go 1.19.
</p>
<h3 id="minor_library_changes">Minor changes to the library</h3>
<p>
As always, there are various minor changes and updates to the library,
made with the Go 1 <a href="/doc/go1compat">promise of compatibility</a>
in mind.
</p>
<dl id="bufio"><dt><a href="/pkg/bufio/">bufio</a></dt>
<dd>
<p><!-- CL 345569 -->
The new <a href="/pkg/bufio#Writer.AvailableBuffer"><code>Writer.AvailableBuffer</code></a>
method returns an empty buffer with a possibly non-empty capacity for use
with append-like APIs. After appending, the buffer can be provided to a
succeeding <code>Write</code> call and possibly avoid any copying.
</p>
<p><!-- CL 345570 -->
The methods <a href="/pkg/bufio#Reader.Reset"><code>Reader.Reset</code></a> and
<a href="/pkg/bufio#Writer.Reset"><code>Writer.Reset</code></a>
now use the default buffer size when called on objects with a
<code>nil</code> buffer.
</p>
</dd>
</dl><!-- bufio -->
<dl id="bytes"><dt><a href="/pkg/bytes/">bytes</a></dt>
<dd>
<p><!-- CL 351710 -->
The new <a href="/pkg/bytes/#Cut"><code>Cut</code></a> function
slices a <code>[]byte</code> around a separator. It can replace
and simplify many common uses of
<a href="/pkg/bytes/#Index"><code>Index</code></a>,
<a href="/pkg/bytes/#IndexByte"><code>IndexByte</code></a>,
<a href="/pkg/bytes/#IndexRune"><code>IndexRune</code></a>,
and <a href="/pkg/bytes/#SplitN"><code>SplitN</code></a>.
</p>
<p><!-- CL 323318, CL 332771 -->
<a href="/pkg/bytes/#Trim"><code>Trim</code></a>, <a href="/pkg/bytes/#TrimLeft"><code>TrimLeft</code></a>,
and <a href="/pkg/bytes/#TrimRight"><code>TrimRight</code></a> are now allocation free and, especially for
small ASCII cutsets, up to 10 times faster.
</p>
<p><!-- CL 359485 -->
The <a href="/pkg/bytes/#Title"><code>Title</code></a> function is now deprecated. It doesn't
handle Unicode punctuation and language-specific capitalization rules, and is superseded by the
<a href="https://golang.org/x/text/cases">golang.org/x/text/cases</a> package.
</p>
</dd>
</dl><!-- bytes -->
<dl id="crypto/elliptic"><dt><a href="/pkg/crypto/elliptic/">crypto/elliptic</a></dt>
<dd>
<p><!-- CL 320071, CL 320072, CL 320074, CL 361402, CL 360014 -->
The <a href="/pkg/crypto/elliptic#P224"><code>P224</code></a>,
<a href="/pkg/crypto/elliptic#P384"><code>P384</code></a>, and
<a href="/pkg/crypto/elliptic#P521"><code>P521</code></a> curve
implementations are now all backed by code generated by the
<a href="https://github.com/mmcloughlin/addchain">addchain</a> and
<a href="https://github.com/mit-plv/fiat-crypto">fiat-crypto</a>
projects, the latter of which is based on a formally-verified model
of the arithmetic operations. They now use safer complete formulas
and internal APIs. P-224 and P-384 are now approximately four times
faster. All specific curve implementations are now constant-time.
</p>
<p>
Operating on invalid curve points (those for which the
<code>IsOnCurve</code> method returns false, and which are never returned
by <a href="/pkg/crypto/elliptic#Unmarshal"><code>Unmarshal</code></a> or
a <code>Curve</code> method operating on a valid point) has always been
undefined behavior, can lead to key recovery attacks, and is now
unsupported by the new backend. If an invalid point is supplied to a
<code>P224</code>, <code>P384</code>, or <code>P521</code> method, that
method will now return a random point. The behavior might change to an
explicit panic in a future release.
</p>
</dd>
</dl><!-- crypto/elliptic -->
<dl id="crypto/tls"><dt><a href="/pkg/crypto/tls/">crypto/tls</a></dt>
<dd>
<p><!-- CL 325250 -->
The new <a href="/pkg/crypto/tls/#Conn.NetConn"><code>Conn.NetConn</code></a>
method allows access to the underlying
<a href="/pkg/net#Conn"><code>net.Conn</code></a>.
</p>
</dd>
</dl><!-- crypto/tls -->
<dl id="crypto/x509"><dt><a href="/pkg/crypto/x509">crypto/x509</a></dt>
<dd>
<p><!-- CL 353132, CL 353403 -->
<a href="/pkg/crypto/x509/#Certificate.Verify"><code>Certificate.Verify</code></a>
now uses platform APIs to verify certificate validity on macOS and iOS when it
is called with a nil
<a href="/pkg/crypto/x509/#VerifyOpts.Roots"><code>VerifyOpts.Roots</code></a>
or when using the root pool returned from
<a href="/pkg/crypto/x509/#SystemCertPool"><code>SystemCertPool</code></a>.
</p>
<p><!-- CL 353589 -->
<a href="/pkg/crypto/x509/#SystemCertPool"><code>SystemCertPool</code></a>
is now available on Windows.
</p>
<p>
On Windows, macOS, and iOS, when a
<a href="/pkg/crypto/x509/#CertPool"><code>CertPool</code></a> returned by
<a href="/pkg/crypto/x509/#SystemCertPool"><code>SystemCertPool</code></a>
has additional certificates added to it,
<a href="/pkg/crypto/x509/#Certificate.Verify"><code>Certificate.Verify</code></a>
will do two verifications: one using the platform verifier APIs and the
system roots, and one using the Go verifier and the additional roots.
Chains returned by the platform verifier APIs will be prioritized.
</p>
<p>
<a href="/pkg/crypto/x509/#CertPool.Subjects"><code>CertPool.Subjects</code></a>
is deprecated. On Windows, macOS, and iOS the
<a href="/pkg/crypto/x509/#CertPool"><code>CertPool</code></a> returned by
<a href="/pkg/crypto/x509/#SystemCertPool"><code>SystemCertPool</code></a>
will return a pool which does not include system roots in the slice
returned by <code>Subjects</code>, as a static list can't appropriately
represent the platform policies and might not be available at all from the
platform APIs.
</p>
</dd>
</dl>
<dl id="debug/dwarf"><dt><a href="/pkg/debug/dwarf/">debug/dwarf</a></dt>
<dd>
<p><!-- CL 380714 -->
The <a href="/pkg/debug/dwarf#StructField"><code>StructField</code></a>
and <a href="/pkg/debug/dwarf#BasicType"><code>BasicType</code></a>
structs both now have a <code>DataBitOffset</code> field, which
holds the value of the <code>DW_AT_data_bit_offset</code>
attribute if present.
</dd>
</dl>
<dl id="debug/elf"><dt><a href="/pkg/debug/elf/">debug/elf</a></dt>
<dd>
<p><!-- CL 352829 -->
The <a href="/pkg/debug/elf/#R_PPC64_RELATIVE"><code>R_PPC64_RELATIVE</code></a>
constant has been added.
</p>
</dd>
</dl><!-- debug/elf -->
<dl id="debug/plan9obj"><dt><a href="/pkg/debug/plan9obj/">debug/plan9obj</a></dt>
<dd>
<p><!-- CL 350229 -->
The <a href="/pkg/debug/plan9obj#File.Symbols">File.Symbols</a>
method now returns the new exported error
value <a href="/pkg/debug/plan9obj#ErrNoSymbols">ErrNoSymbols</a>
if the file has no symbol section.
</p>
</dd>
</dl><!-- debug/plan9obj -->
<dl id="go/ast"><dt><a href="/pkg/go/ast/">go/ast</a></dt>
<dd>
<p><!-- https://golang.org/issue/47781, CL 325689, CL 327149, CL 348375, CL 348609 -->
Per the proposal
<a href="https://go.googlesource.com/proposal/+/master/design/47781-parameterized-go-ast.md">
Additions to go/ast and go/token to support parameterized functions and types
</a>
the following additions are made to the <a href="/pkg/go/ast"><code>go/ast</code></a> package:
<ul>
<li>
the <a href="/pkg/go/ast/#FuncType"><code>FuncType</code></a>
and <a href="/pkg/go/ast/#TypeSpec"><code>TypeSpec</code></a>
nodes have a new field <code>TypeParams</code> to hold type parameters, if any.
</li>
<li>
The new expression node <a href="/pkg/go/ast/#IndexListExpr"><code>IndexListExpr</code></a>
represents index expressions with multiple indices, used for function and type instantiations
with more than one explicit type argument.
</li>
</ul>
</p>
</dd>
</dl>
<dl id="go/constant"><dt><a href="/pkg/go/constant/">go/constant</a></dt>
<dd>
<p><!-- https://golang.org/issue/46211, CL 320491 -->
The new <a href="/pkg/go/constant/#Kind.String"><code>Kind.String</code></a>
method returns a human-readable name for the receiver kind.
</p>
</dd>
</dl>
<dl id="go/token"><dt><a href="/pkg/go/token/">go/token</a></dt>
<dd>
<p><!-- https://golang.org/issue/47781, CL 324992 -->
The new constant <a href="/pkg/go/token/#TILDE"><code>TILDE</code></a>
represents the <code>~</code> token per the proposal
<a href="https://go.googlesource.com/proposal/+/master/design/47781-parameterized-go-ast.md">
Additions to go/ast and go/token to support parameterized functions and types
</a>.
</p>
</dd>
</dl>
<dl id="go/types"><dt><a href="/pkg/go/types/">go/types</a></dt>
<dd>
<p><!-- https://golang.org/issue/46648 -->
The new <a href="/pkg/go/types/#Config.GoVersion"><code>Config.GoVersion</code></a>
field sets the accepted Go language version.
</p>
<p><!-- https://golang.org/issue/47916 -->
Per the proposal
<a href="https://go.googlesource.com/proposal/+/master/design/47916-parameterized-go-types.md">
Additions to go/types to support type parameters
</a>
the following additions are made to the <a href="/pkg/go/types"><code>go/types</code></a> package:
</p>
<ul>
<li>
The new type
<a href="/pkg/go/types/#TypeParam"><code>TypeParam</code></a>, factory function
<a href="/pkg/go/types/#NewTypeParam"><code>NewTypeParam</code></a>,
and associated methods are added to represent a type parameter.
</li>
<li>
The new type
<a href="/pkg/go/types/#TypeParamList"><code>TypeParamList</code></a> holds a list of
type parameters.
</li>
<li>
The new type
<a href="/pkg/go/types/#TypeList"><code>TypeList</code></a> holds a list of types.
</li>
<li>
The new factory function
<a href="/pkg/go/types/#NewSignatureType"><code>NewSignatureType</code></a> allocates a
<a href="/pkg/go/types/#Signature"><code>Signature</code></a> with
(receiver or function) type parameters.
To access those type parameters, the <code>Signature</code> type has two new methods
<a href="/pkg/go/types/#Signature.RecvTypeParams"><code>Signature.RecvTypeParams</code></a> and
<a href="/pkg/go/types/#Signature.TypeParams"><code>Signature.TypeParams</code></a>.
</li>
<li>
<a href="/pkg/go/types/#Named"><code>Named</code></a> types have four new methods:
<a href="/pkg/go/types/#Named.Origin"><code>Named.Origin</code></a> to get the original
parameterized types of instantiated types,
<a href="/pkg/go/types/#Named.TypeArgs"><code>Named.TypeArgs</code></a> and
<a href="/pkg/go/types/#Named.TypeParams"><code>Named.TypeParams</code></a> to get the
type arguments or type parameters of an instantiated or parameterized type, and
<a href="/pkg/go/types/#Named.TypeParams"><code>Named.SetTypeParams</code></a> to set the
type parameters (for instance, when importing a named type where allocation of the named
type and setting of type parameters cannot be done simultaneously due to possible cycles).
</li>
<li>
The <a href="/pkg/go/types/#Interface"><code>Interface</code></a> type has four new methods:
<a href="/pkg/go/types/#Interface.IsComparable"><code>Interface.IsComparable</code></a> and
<a href="/pkg/go/types/#Interface.IsMethodSet"><code>Interface.IsMethodSet</code></a> to
query properties of the type set defined by the interface, and
<a href="/pkg/go/types/#Interface.MarkImplicit"><code>Interface.MarkImplicit</code></a> and
<a href="/pkg/go/types/#Interface.IsImplicit"><code>Interface.IsImplicit</code></a> to set
and test whether the interface is an implicit interface around a type constraint literal.
</li>
<li>
The new types
<a href="/pkg/go/types/#Union"><code>Union</code></a> and
<a href="/pkg/go/types/#Term"><code>Term</code></a>, factory functions
<a href="/pkg/go/types/#NewUnion"><code>NewUnion</code></a> and
<a href="/pkg/go/types/#NewTerm"><code>NewTerm</code></a>, and associated
methods are added to represent type sets in interfaces.
</li>
<li>
The new function
<a href="/pkg/go/types/#Instantiate"><code>Instantiate</code></a>
instantiates a parameterized type.
</li>
<li>
The new <a href="/pkg/go/types/#Info.Instances"><code>Info.Instances</code></a>
map records function and type instantiations through the new
<a href="/pkg/go/types/#Instance"><code>Instance</code></a> type.
</li>
<li><!-- CL 342671 -->
The new type <a href="/pkg/go/types/#ArgumentError"><code>ArgumentError</code></a>
and associated methods are added to represent an error related to a type argument.
</li>
<li><!-- CL 353089 -->
The new type <a href="/pkg/go/types/#Context"><code>Context</code></a> and factory function
<a href="/pkg/go/types/#NewContext"><code>NewContext</code></a>
are added to facilitate sharing of identical type instances
across type-checked packages, via the new
<a href="/pkg/go/types/#Config.Context"><code>Config.Context</code></a>
field.
</li>
</ul>
</dd>
</dl>
<dl id="html/template"><dt><a href="/pkg/html/template/">html/template</a></dt>
<dd>
<p><!-- CL 321491 -->
Within a <code>range</code> pipeline the new
<code>{{break}}</code> command will end the loop early and the
new <code>{{continue}}</code> command will immediately start the
next loop iteration.
</p>
<p><!-- CL 321490 -->
The <code>and</code> function no longer always evaluates all arguments; it
stops evaluating arguments after the first argument that evaluates to
false. Similarly, the <code>or</code> function now stops evaluating
arguments after the first argument that evaluates to true. This makes a
difference if any of the arguments is a function call.
</p>
</dd>
</dl><!-- html/template -->
<dl id="image/draw"><dt><a href="/pkg/image/draw/">image/draw</a></dt>
<dd>
<p><!-- CL 340049 -->
The <code>Draw</code> and <code>DrawMask</code> fallback implementations
(used when the arguments are not the most common image types) are now
faster when those arguments implement the optional
<a href="/pkg/image/draw/#RGBA64Image"><code>draw.RGBA64Image</code></a>
and <a href="/pkg/image/#RGBA64Image"><code>image.RGBA64Image</code></a>
interfaces that were added in Go 1.17.
</p>
</dd>
</dl><!-- image/draw -->
<dl id="net"><dt><a href="/pkg/net/">net</a></dt>
<dd>
<p><!-- CL 340261 -->
<a href="/pkg/net#Error"><code>net.Error.Temporary</code></a> has been deprecated.
</p>
</dd>
</dl><!-- net -->
<dl id="net/http"><dt><a href="/pkg/net/http/">net/http</a></dt>
<dd>
<p><!-- CL 330852 -->
On WebAssembly targets, the <code>Dial</code>, <code>DialContext</code>,
<code>DialTLS</code> and <code>DialTLSContext</code> method fields in
<a href="/pkg/net/http/#Transport"><code>Transport</code></a>
will now be correctly used, if specified, for making HTTP requests.
</p>
<p><!-- CL 338590 -->
The new
<a href="/pkg/net/http#Cookie.Valid"><code>Cookie.Valid</code></a>
method reports whether the cookie is valid.
</p>
<p><!-- CL 346569 -->
The new
<a href="/pkg/net/http#MaxBytesHandler"><code>MaxBytesHandler</code></a>
function creates a <code>Handler</code> that wraps its
<code>ResponseWriter</code> and <code>Request.Body</code> with a
<a href="/pkg/net/http#MaxBytesReader"><code>MaxBytesReader</code></a>.
</p>
</dd>
</dl><!-- net/http -->
<dl id="os/user"><dt><a href="/pkg/os/user/">os/user</a></dt>
<dd>
<p><!-- CL 330753 -->
<a href="/pkg/os/user#User.GroupIds"><code>User.GroupIds</code></a>.
now uses a Go native implementation when cgo is not available.
</p>
</dd>
</dl><!-- os/user -->
<dl id="reflect"><dt><a href="/pkg/reflect/">reflect</a></dt>
<dd>
<p><!-- CL 356049, CL 320929 -->
The new
<a href="/pkg/reflect/#Value.SetIterKey"><code>Value.SetIterKey</code></a>
and <a href="/pkg/reflect/#Value.SetIterValue"><code>Value.SetIterValue</code></a>
methods set a Value using a map iterator as the source. They are equivalent to
<code>Value.Set(iter.Key())</code> and <code>Value.Set(iter.Value())</code> but
do fewer allocations.
</p>
<p><!-- CL 350691 -->
The new
<a href="/pkg/reflect/#Value.UnsafePointer"><code>Value.UnsafePointer</code></a>
method returns the Value's value as an <a href="/pkg/unsafe/#Pointer"><code>unsafe.Pointer</code></a>.
This allows callers to migrate from <a href="/pkg/reflect/#Value.UnsafeAddr"><code>Value.UnsafeAddr</code></a>
and <a href="/pkg/reflect/#Value.Pointer"><code>Value.Pointer</code></a>
to eliminate the need to perform uintptr to unsafe.Pointer conversions at the callsite (as unsafe.Pointer rules require).
</p>
<p><!-- CL 321891 -->
The new
<a href="/pkg/reflect/#MapIter.Reset"><code>MapIter.Reset</code></a>
method changes its receiver to iterate over a
different map. The use of
<a href="/pkg/reflect/#MapIter.Reset"><code>MapIter.Reset</code></a>
allows allocation-free iteration
over many maps.
</p>
<p><!-- CL 352131 -->
A number of methods (
<a href="/pkg/reflect#Value.CanInt"><code>Value.CanInt</code></a>,
<a href="/pkg/reflect#Value.CanUint"><code>Value.CanUint</code></a>,
<a href="/pkg/reflect#Value.CanFloat"><code>Value.CanFloat</code></a>,
<a href="/pkg/reflect#Value.CanComplex"><code>Value.CanComplex</code></a>
)
have been added to
<a href="/pkg/reflect#Value"><code>Value</code></a>
to test if a conversion is safe.
</p>
<p><!-- CL 357962 -->
<a href="/pkg/reflect#Value.FieldByIndexErr"><code>Value.FieldByIndexErr</code></a>
has been added to avoid the panic that occurs in
<a href="/pkg/reflect#Value.FieldByIndex"><code>Value.FieldByIndex</code></a>
when stepping through a nil pointer to an embedded struct.
</p>
<p><!-- CL 341333 -->
<a href="/pkg/reflect#Ptr"><code>reflect.Ptr</code></a> and
<a href="/pkg/reflect#PtrTo"><code>reflect.PtrTo</code></a>
have been renamed to
<a href="/pkg/reflect#Pointer"><code>reflect.Pointer</code></a> and
<a href="/pkg/reflect#PointerTo"><code>reflect.PointerTo</code></a>,
respectively, for consistency with the rest of the reflect package.
The old names will continue to work, but will be deprecated in a
future Go release.
</p>
</dd>
</dl><!-- reflect -->
<dl id="regexp"><dt><a href="/pkg/regexp/">regexp</a></dt>
<dd>
<p><!-- CL 354569 -->
<a href="/pkg/regexp/"><code>regexp</code></a>
now treats each invalid byte of a UTF-8 string as <code>U+FFFD</code>.
</p>
</dd>
</dl><!-- regexp -->
<dl id="runtime/debug"><dt><a href="/pkg/runtime/debug/">runtime/debug</a></dt>
<dd>
<p><!-- CL 354569 -->
The <a href="/pkg/runtime/debug#BuildInfo"><code>BuildInfo</code></a>
struct has two new fields, containing additional information
about how the binary was built:
<ul>
<li><a href="/pkg/runtime/debug#BuildInfo.GoVersion"><code>GoVersion</code></a>
holds the version of Go used to build the binary.
</li>
<li>
<a href="/pkg/runtime/debug#BuildInfo.Settings"><code>Settings</code></a>
is a slice of
<a href="/pkg/runtime/debug#BuildSettings"><code>BuildSettings</code></a>
structs holding key/value pairs describing the build.
</li>
</ul>
</p>
</dd>
</dl><!-- runtime/debug -->
<dl id="strconv"><dt><a href="/pkg/strconv/">strconv</a></dt>
<dd>
<p><!-- CL 343877 -->
<a href="/pkg/strconv/#strconv.Unquote"><code>strconv.Unquote</code></a>
now rejects Unicode surrogate halves.
</p>
</dd>
</dl><!-- strconv -->
<dl id="strings"><dt><a href="/pkg/strings/">strings</a></dt>
<dd>
<p><!-- CL 351710 -->
The new <a href="/pkg/strings/#Cut"><code>Cut</code></a> function
slices a <code>string</code> around a separator. It can replace
and simplify many common uses of
<a href="/pkg/strings/#Index"><code>Index</code></a>,
<a href="/pkg/strings/#IndexByte"><code>IndexByte</code></a>,
<a href="/pkg/strings/#IndexRune"><code>IndexRune</code></a>,
and <a href="/pkg/strings/#SplitN"><code>SplitN</code></a>.
</p>
<p><!-- CL 345849 -->
The new <a href="/pkg/strings/#Clone"><code>Clone</code></a> function copies the input
<code>string</code> without the returned cloned <code>string</code> referencing
the input string's memory.
</p>
<p><!-- CL 323318, CL 332771 -->
<a href="/pkg/strings/#Trim"><code>Trim</code></a>, <a href="/pkg/strings/#TrimLeft"><code>TrimLeft</code></a>,
and <a href="/pkg/strings/#TrimRight"><code>TrimRight</code></a> are now allocation free and, especially for
small ASCII cutsets, up to 10 times faster.
</p>
<p><!-- CL 359485 -->
The <a href="/pkg/strings/#Title"><code>Title</code></a> function is now deprecated. It doesn't
handle Unicode punctuation and language-specific capitalization rules, and is superseded by the
<a href="https://golang.org/x/text/cases">golang.org/x/text/cases</a> package.
</p>
</dd>
</dl><!-- strings -->
<dl id="sync"><dt><a href="/pkg/sync/">sync</a></dt>
<dd>
<p><!-- CL 319769 -->
The new methods
<a href="/pkg/sync#Mutex.TryLock"><code>Mutex.TryLock</code></a>,
<a href="/pkg/sync#RWMutex.TryLock"><code>RWMutex.TryLock</code></a>, and
<a href="/pkg/sync#RWMutex.TryRLock"><code>RWMutex.TryRLock</code></a>,
will acquire the lock if it is not currently held.
</p>
</dd>
</dl><!-- sync -->
<dl id="syscall"><dt><a href="/pkg/syscall/">syscall</a></dt>
<dd>
<p><!-- CL 336550 -->
The new function <a href="/pkg/syscall/?GOOS=windows#SyscallN"><code>SyscallN</code></a>
has been introduced for Windows, allowing for calls with arbitrary number
of arguments. As a result,
<a href="/pkg/syscall/?GOOS=windows#Syscall"><code>Syscall</code></a>,
<a href="/pkg/syscall/?GOOS=windows#Syscall6"><code>Syscall6</code></a>,
<a href="/pkg/syscall/?GOOS=windows#Syscall9"><code>Syscall9</code></a>,
<a href="/pkg/syscall/?GOOS=windows#Syscall12"><code>Syscall12</code></a>,
<a href="/pkg/syscall/?GOOS=windows#Syscall15"><code>Syscall15</code></a>, and
<a href="/pkg/syscall/?GOOS=windows#Syscall18"><code>Syscall18</code></a> are
deprecated in favor of <a href="/pkg/syscall/?GOOS=windows#SyscallN"><code>SyscallN</code></a>.
</p>
<p><!-- CL 355570 -->
<a href="/pkg/syscall/?GOOS=freebsd#SysProcAttr.Pdeathsig"><code>SysProcAttr.Pdeathsig</code></a>.
is now supported in FreeBSD.
</p>
</dd>
cmd/gofmt: format files in parallel gofmt is pretty heavily CPU-bound, since parsing and formatting 1MiB of Go code takes much longer than reading that amount of bytes from disk. However, parsing and manipulating a large Go source file is very difficult to parallelize, so we continue to process each file in its own goroutine. A Go module may contain a large number of Go source files, so we need to bound the amount of work in flight. However, because the distribution of sizes for Go source files varies widely — from tiny doc.go files containing a single package comment all the way up to massive API wrappers generated by automated tools — the amount of time, work, and memory overhead needed to process each file also varies. To account for this variability, we limit the in-flight work by bytes of input rather than by number of files. That allows us to make progress on many small files while we wait for work on a handful of large files to complete. The gofmt tool has a well-defined output format on stdout, which was previously deterministic. We keep it deterministic by printing the results of each file in order, using a lazily-synchronized io.Writer (loosly inspired by Haskell's IO monad). After a file has been formatted in memory, we keep it in memory (again, limited by the corresponding number of input bytes) until the output for all previous files has been flushed. This adds a bit of latency compared to emitting the output in nondeterministic order, but a little extra latency seems worth the cost to preserve output stability. This change is based on Daniel Martí's work in CL 284139, but using a weighted semaphore and ephemeral goroutines instead of a worker pool and batches. Benchmark results are similar, and I find the concurrency in this approach a bit easier to reason about. In the batching-based approach, the batch size allows us to "look ahead" to find large files and start processing them early. To keep the CPUs saturated and prevent stragglers, we would need to tune the batch size to be about the same as the largest input files. If the batch size is set too high, a large batch of small files could turn into a straggler, but if the batch size is set too low, the largest files in the data set won't be started early enough and we'll end up with a large-file straggler. One possible alternative would be to sort by file size instead of batching: identify all of the files to be processed, sort from largest to smallest, and then process the largest files first so that the "tail" of processing covers the smallest files. However, that approach would still fail to saturate available CPU when disk latency is high, would require buffering an arbitrary amount of metadata in order to sort by size, and (perhaps most importantly!) would not allow the `gofmt` binary to preserve the same (deterministic) output order that it has today. In contrast, with a semaphore we can produce the same deterministic output as ever using only one tuning parameter: the memory footprint, expressed as a rough lower bound on the amount of RAM available per thread. While we're below the memory limit, we can run arbitrarily many disk operations arbitrarily far ahead, and process the results of those operations whenever they become avaliable. Then it's up to the kernel (not us) to schedule the disk operations for throughput and latency, and it's up to the runtime (not us) to schedule the goroutines so that they complete quickly. In practice, even a modest assumption of a few megabytes per thread seems to provide a nice speedup, and it should scale reasonably even to machines with vastly different ratios of CPU to disk. (In practice, I expect that most 'gofmt' invocations will work with files on at most one physical disk, so the CPU:disk ratio should vary more-or-less directly with the thread count, whereas the CPU:memory ratio is more-or-less independent of thread count.) name \ time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 11.9s ± 2% 2.7s ± 3% 2.8s ± 5% name \ user-time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 13.5s ± 2% 14.4s ± 1% 14.7s ± 1% name \ sys-time/op baseline.txt 284139.txt simplified.txt GofmtGorootCmd 465ms ± 8% 229ms ±28% 232ms ±31% name \ peak-RSS-bytes baseline.txt 284139.txt simplified.txt GofmtGorootCmd 77.7MB ± 4% 162.2MB ±10% 192.9MB ±15% For #43566 Change-Id: I4ba251eb4d188a3bd1901039086be57f0b341910 Reviewed-on: https://go-review.googlesource.com/c/go/+/317975 Trust: Bryan C. Mills <bcmills@google.com> Trust: Daniel Martí <mvdan@mvdan.cc> Run-TryBot: Bryan C. Mills <bcmills@google.com> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
2021-01-16 11:03:31 -07:00
</dl><!-- syscall -->
<dl id="syscall/js"><dt><a href="/pkg/syscall/js/">syscall/js</a></dt>
<dd>
<p><!-- CL 356430 -->
<code>Wrapper</code> interface has been removed.
</p>
</dd>
</dl><!-- syscall/js -->
<dl id="testing"><dt><a href="/pkg/testing/">testing</a></dt>
<dd>
<p><!-- CL 343883 -->
The precedence of <code>/</code> in the argument for <code>-run</code> and
<code>-bench</code> has been increased. <code>A/B|C/D</code> used to be
treated as <code>A/(B|C)/D</code> and is now treated as
<code>(A/B)|(C/D)</code>.
</p>
<p><!-- CL 356669 -->
If the <code>-run</code> option does not select any tests, the
<code>-count</code> option is ignored. This could change the behavior of
existing tests in the unlikely case that a test changes the set of subtests
that are run each time the test function itself is run.
</p>
<p><!-- CL 251441 -->
The new <a href="/pkg/testing#F"><code>testing.F</code></a> type
is used by the new <a href="#fuzzing">fuzzing support described
above</a>. Tests also now support the command line
options <code>-test.fuzz</code>, <code>-test.fuzztime</code>, and
<code>-test.fuzzminimizetime</code>.
</p>
</dd>
</dl><!-- testing -->
<dl id="text/template"><dt><a href="/pkg/text/template/">text/template</a></dt>
<dd>
<p><!-- CL 321491 -->
Within a <code>range</code> pipeline the new
<code>{{break}}</code> command will end the loop early and the
new <code>{{continue}}</code> command will immediately start the
next loop iteration.
</p>
<p><!-- CL 321490 -->
The <code>and</code> function no longer always evaluates all arguments; it
stops evaluating arguments after the first argument that evaluates to
false. Similarly, the <code>or</code> function now stops evaluating
arguments after the first argument that evaluates to true. This makes a
difference if any of the arguments is a function call.
</p>
</dd>
</dl><!-- text/template -->
<dl id="text/template/parse"><dt><a href="/pkg/text/template/parse/">text/template/parse</a></dt>
<dd>
<p><!-- CL 321491 -->
The package supports the new
<a href="/pkg/text/template/">text/template</a> and
<a href="/pkg/html/template/">html/template</a>
<code>{{break}}</code> command via the new constant
<a href="/pkg/text/template/parse#NodeBreak"><code>NodeBreak</code></a>
and the new type
<a href="/pkg/text/template/parse#BreakNode"><code>BreakNode</code></a>,
and similarly supports the new <code>{{continue}}</code> command
via the new constant
<a href="/pkg/text/template/parse#NodeContinue"><code>NodeContinue</code></a>
and the new type
<a href="/pkg/text/template/parse#ContinueNode"><code>ContinueNode</code></a>.
</p>
</dd>
</dl><!-- text/template -->
<dl id="unicode/utf8"><dt><a href="/pkg/unicode/utf8/">unicode/utf8</a></dt>
<dd>
<p><!-- CL 345571 -->
The <a href="/pkg/unicode/utf8/#AppendRune"><code>AppendRune</code></a> function appends the UTF-8 new
encoding of a <code>rune</code> to a <code>[]byte</code>.
</p>
</dd>
</dl><!-- unicode/utf8 -->