path: root/doc/go1.1.html

<!--{
	"Title": "Go 1.1 Release Notes",
	"Path":  "/doc/go1.1",
	"Template": true
}-->

<h2 id="introduction">Introduction to Go 1.1</h2>

TODO
 - overview
 - link back to Go 1 and also Go 1 Compatibility docs.

<h2 id="language">Changes to the language</h2>

TODO

<h3 id="divzero">Integer division by zero</h3>

<p>
In Go 1, integer division by a constant zero produced a runtime panic:
</p>

<pre>
func f(x int) int {
	return x/0
}
</pre>

<p>
In Go 1.1, an integer division by constant zero is not a legal program, so it is a compile-time error.
</p>


<h2 id="impl">Changes to the implementations and tools</h2>

TODO: more

<h3 id="gc-flag">Command-line flag parsing</h3>

<p>
In the gc toolchain, the compilers and linkers now use the
same command-line flag parsing rules as the Go flag package, a departure
from the traditional Unix flag parsing. This may affect scripts that invoke
the tool directly.
For example,
<code>go tool 6c -Fw -Dfoo</code> must now be written
<code>go tool 6c -F -w -D foo</code>. 
</p>

<h3 id="int">Size of int on 64-bit platforms</h3>

<p>
The language allows the implementation to choose whether the <code>int</code> type and <code>uint</code> types are 32 or 64 bits. Previous Go implementations made <code>int</code> and <code>uint</code> 32 bits on all systems. Both the gc and gccgo implementations (TODO: check that gccgo does) <a href="http://golang.org/issue/2188">now make <code>int</code> and <code>uint</code> 64 bits on 64-bit platforms such as AMD64/x86-64</a>.
Among other things, this enables the allocation of slices with
more than 2 billion elements on 64-bit platforms.
</p>

<p>
<em>Updating</em>:
Most programs will be unaffected by this change.
Because Go does not allow implicit conversions between distinct
<a href="/ref/spec#Numeric_types">numeric types</a>,
no programs will stop compiling due to this change.
However, programs that contain implicit assumptions
that <code>int</code> is only 32 bits may change behavior.
For example, this code prints a positive number on 64-bit systems and
a negative one on 32-bit systems:

<pre>
x := ^uint32(0) // x is 0xffffffff
i := int(x)     // i is -1 on 32-bit systems, 0xffffffff on 64-bit
fmt.Println(i)
</pre>

<p>Portable code intending 32-bit sign extension (yielding -1 on all systems)
would instead say:
</p>

<pre>
i := int(int32(x))
</pre>

<h3 id="asm">Assembler</h3>

<p>
Due to the <a href="#int">int</a> and TODO: OTHER changes,
the placement of function arguments on the stack has changed.
Functions written in assembly will need to be revised at least
to adjust frame pointer offsets.
</p>

<h3 id="gotool">Changes to the go tool</h3>

<p>The <code>go</code> tool has acquired several improvements which are intended to improve the experience for new Go users.</p>

<p>Firstly, when compiling, testing, or running Go code, the <code>go</code> tool will now give more detailed errors messages, including a list of paths searched, when a package cannot be located.
</p>

<pre>
$ go build foo/quxx
can't load package: package foo/quxx: cannot find package "foo/quxx" in any of:
        /home/User/go/src/pkg/foo/quxx (from $GOROOT)
        /home/User/src/foo/quxx (from $GOPATH) 
</pre>

<p>
Secondly, the <code>go get</code> command no longer allows <code>$GOROOT</code> as the default destination when downloading package source. To use <code>go get</code> command, a valid <code>$GOPATH</code> is now required.
</p>

<pre>
$ GOPATH= go get code.google.com/p/foo/quxx
package code.google.com/p/foo/quxx: cannot download, $GOPATH not set. For more details see: go help gopath 
</pre>

<p>Finally, as a result of the previous change, the <code>go get</code> command will also fail when <code>$GOPATH</code> and <code>$GOROOT</code> are set to the same value. 
</p>

<pre>
$ GOPATH=$GOROOT go get code.google.com/p/foo/quxx
warning: GOPATH set to GOROOT (/home/User/go) has no effect
package code.google.com/p/foo/quxx: cannot download, $GOPATH must not be set to $GOROOT. For more details see: go help gopath
</pre>

<h3 id="gofix">Changes to go fix</h3>

<p>
The <code>go fix</code> command no longer applies fixes to update code from
before Go 1 to use Go 1 APIs. To update pre-Go 1 code to Go 1.1, use a Go 1.0 toolchain
to convert the code to Go 1.0 first.
</p>

<h2 id="library">Changes to the standard library</h2>

<h3 id="debug_elf">debug/elf</h3>
<p>
Previous versions of the debug/elf package intentionally skipped over the first
symbol in the ELF symbol table, since it is always an empty symbol. This symbol
is no longer skipped since indexes into the symbol table returned by debug/elf,
will be different to indexes into the original ELF symbol table. Any code that
calls the debug/elf functions Symbols or ImportedSymbols may need to be
adjusted to account for the additional symbol and the change in symbol offsets.
</p>

<h3 id="html_template">html/template</h3>

<p>
Templates using the undocumented and only partially implemented
"noescape" feature will break: that feature was removed.
</p>

<h3 id="net">net</h3>

<p>
The protocol-specific resolvers were formerly
lax about the network name passed in. For example, although the documentation was clear
that the only valid networks for <code>ResolveTCPAddr</code> are <code>"tcp"</code>,
<code>"tcp4"</code>, and <code>"tcp6"</code>, the Go 1.0 implementation silently accepted
any string. The Go 1.1 implementation returns an error if the network is not one of those strings.
The same is true of the other protocol-specific resolvers <code>ResolveIPAddr</code>, <code>ResolveUDPAddr</code>, and
<code>ResolveUnixAddr</code>.
</p>

<p>
The previous <code>ListenUnixgram</code> returned <code>UDPConn</code> as
arepresentation of the connection endpoint. The Go 1.1 implementation
returns <code>UnixConn</code> to allow reading and writing
with <code>ReadFrom</code> and <code>WriteTo</code> methods on
the <code>UnixConn</code>.
</p>

<h3 id="time">time</h3>
<p>
On FreeBSD, Linux, NetBSD, OS X and OpenBSD, previous versions of the time package
returned times with microsecond precision. The Go 1.1 implementation of time on these
systems now returns times with nanosecond precision. Code may exist that expects to be
able to store such a time in an external format with only microsecond precision,
read it back, and recover exactly the same time instant.
In Go 1.1 the same time will not be recovered, since the external storage
will have discarded nanoseconds.
To address this case, there are two new methods of time.Time, Round and Truncate,
that can be used to remove precision from a time before passing it to
external storage.
</p>

<h3 id="exp_old">Exp and old subtrees moved to go.exp subrepo</h3>

<p>
To make it easier for binary distributions to access them if desired, the <code>exp</code>
and <code>old</code> source subtrees, which are not included in binary distributions,
have been moved to the new <code>go.exp</code> subrepository at
<code>code.google.com/p/go.exp</code>. To access the <code>ssa</code> package,
for example, run
</p>

<pre>
$ go get code.google.com/p/go.exp/ssa
</pre>

<p>
and then in Go source,
</p>

<pre>
import "code.google.com/p/go.exp/ssa"
</pre>

<h3 id="TODO">TODO</h3>