This is documenting the status quo. The previous cleanup
added this language as an implementation restriction, but
at least for now it is really part of the language proper.
Fixes#4605.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/7305071
The gc compilers already behave this way. It was an explicit
decision during the very long constant design discussions,
but it appears not to have made it into the spec.
Fixes#4398.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/7313069
The spec is not clear about whether this is allowed or not,
but both compilers allow it, because the reflect implementation
takes advantage of it. Document current behavior.
Fixes#4679.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/7303064
The spec talks explicitly about rune literals but the
respective production is still called char_lit for
historic reasons. Updated the two occurences.
Fixes#4602.
R=rsc, iant, r, ken
CC=golang-dev
https://golang.org/cl/7070048
A composite literal may be parenthesized when
used as operand for the unary operator &.
R=rsc, iant, r, ken
CC=golang-dev
https://golang.org/cl/6996053
This is language change. It is a backward-compatible
change but for code that relies on a run-time panic
when calling delete on a nil map (unlikely).
Fixes#4253.
R=rsc, r, iant, ken, bradfitz, rogpeppe
CC=golang-dev
https://golang.org/cl/6909060
We have been using all three terms "indices", "indexes",
and "index expressions" indiscriminatly for index values.
With this change, "index" refers to an index value,
"indices" is the plural of "index", and "index expression"
refers to an array, slice, or map indexed by an index: a[x].
R=r, rsc, iant, ken, mtj
CC=golang-dev
https://golang.org/cl/6912056
The spec didn't preclude invalid type assertions and
type switches, i.e., cases where a concrete type doesn't
implement the interface type in the assertion in the first
place. Both, the gc and gccgo compiler exclude these cases.
This is documenting the status quo.
Also:
- minor clean up of respective examples
- added sentence about default case in select statements
Fixes#4472.
R=rsc, iant, r, ken
CC=golang-dev
https://golang.org/cl/6869050
Also:
- 'for' statements with a range clause do not accept send-only
channels
- '_, _ = range ch' is not equivalent to "_ = range ch" if ch
is a channel (rewriting the latter to the former leads to
an invalid range clause).
These clarifications document the status quo.
R=rsc, r, iant, ken
CC=golang-dev
https://golang.org/cl/6874053
1) Built-ins are restricted like for expression statements.
This documents the status quo.
2) Calls cannot be parenthesized. The spec is not clear. gccgo
permits it already, gc doesn't. Be explicit in the spec.
Fixes#4462.
R=rsc, iant, r, ken, lvd
CC=golang-dev
https://golang.org/cl/6861043
We have the notion of a PackageName, not package identifier.
As is, it could construed that imports that rename a package
don't have an "imported package identifier" but a local one.
R=r, rsc, iant, ken, dsymonds
CC=golang-dev
https://golang.org/cl/6858049
This is an attempt at making the interaction between
these three constructs clearer. Specifically:
- return statements terminate a function, execute deferred
functions, return to the caller, and then execution
continues after the call
- panic calls terminate a function, execute deferred
functions, return to the caller, and then re-panic
- deferred functions are executed before a function _returns_
to its caller
The hope is that with this change it becomes clear when a
deferred function is executed (when a function returns),
and when it is not (when a program exits).
R=r, rsc, iant, ken, iant
CC=golang-dev
https://golang.org/cl/6736071
Both gc and gccgo always checked this for constant
expressions but the spec only mentions run-time
exceptions.
This CL also requires that constant divisors
must not be zero in non-constant integer expressions:
This is consistent with the spirit of the most
recent changes and it is consistent with constant
expressions. We don't want to specify the effect for
non-integer expressions (f/0.0 where f is a float or
complex number) because there the result f/g is not
further specified if a non-constant g is 0.
R=r, rsc, iant, ken, andybalholm, iant
CC=golang-dev
https://golang.org/cl/6710045
This is a language change: Until now, the spec required run-
time panics for some of these errors. Note that gc and gccgo
implemented this inconsistently, and that both compilers already
reported compile-time errors in some cases. This change makes
make() behave along the same vein as index expressions.
This addresses the spec aspect of issue 4085.
R=r, rsc, iant, ken
CC=golang-dev
https://golang.org/cl/6725053
At the moment, gc and gccgo report compile-
time errors for certain constant indexes that
are out of bounds. The spec however requests
a run-time panic for out-of-bounds indexes
(http://tip.golang.org/ref/spec#Indexes).
Document the status quo.
Fixes#4231.
R=r, rsc, iant, ken
CC=golang-dev
https://golang.org/cl/6699048
Proposed new text to make matters clearer. The existing text was
unclear about the state of result parameters when panicking.
R=golang-dev, iant
CC=golang-dev
https://golang.org/cl/6653047
It's already there but only in the "for instance" and so not
clear enough: deferred functions run after
the result parameters are updated.
R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/6631058
Also: Be explicit what operator means with respect to conversion types.
The parenthesis requirement is a language change. At the moment,
literal function types in conversions that cannot possibly be
followed by a '(' don't need parentheses. For instance:
func(int)int(x) -> same as (func(int)int)(x)
func()()(x) -> same as (func())(x)
but:
func(int)(x) -> could be func(int)x {...}
Fixes#4109.
R=rsc, r, iant, ken, iant
CC=golang-dev
https://golang.org/cl/6584065
Also: Adjust prose in Conversions section
to be consistent.
This is a backward-compatible syntax change.
For a discussion see issue 2568.
Fixes#2568.
Fixes#4123.
R=golang-dev, r, iant, rsc
CC=golang-dev
https://golang.org/cl/6572049
Function and method calls are valid expression statements,
but calling certain built-in functions is not permitted.
Enumerate the built-ins.
Also: unsafe.Offsetof permits parenthesized selectors as
arguments.
This is simply documenting existing compiler behavior
(both gc and gccgo agree).
R=r, rsc, iant, ken
CC=golang-dev
https://golang.org/cl/6498138
Both gc and gccgo permit calls such as unsafe.Sizeof(42). The
spec only permits variable arguments. This is a (backward-compatible)
spec change reflecting the status quo. Seems preferrable over
restricting the compilers.
R=r, rsc, iant, ken
CC=golang-dev
https://golang.org/cl/6494140
Strings happen to be represented similarly to
byte slices internally, but they don't quite
behave like them: While strings can be indexed,
sliced, and have their len() taken like byte
slices, string elements are not addressable,
make() and cap() is not supported, range loops
operate differently, and they are immutable (and
thus behave like values rather then references).
Fixes#4018.
R=r, rsc, iant, ken
CC=golang-dev
https://golang.org/cl/6503116
The (and not) arguably sounds like it is trying to say something - and not what?.
Just an idea, won't be hurt if it gets rejected.
R=gri, dsymonds, r
CC=golang-dev
https://golang.org/cl/6498115
After further deliberation, let's back down to the Unicode proposal.
Ignoring aBOMinations anywhere means that things like
grep unsafe *.go
might fail because there's a BOM in the middle: unBOMsafe.
R=golang-dev, rsc, 0xjnml, gri, bradfitz
CC=golang-dev
https://golang.org/cl/6490091
No changes to the meaning, just clearer language and more
examples, including illegal rune and string literals.
In particular, "character literal" and "character constant"
are now called "rune literal" and "rune constant" and the
word "character" always refers to the source text, not
program values.
R=golang-dev, gri
CC=golang-dev
https://golang.org/cl/6448137
- point out difference between selectors and qualified identifiers
- differentiate between illegal selectors and run-time panics
- use "indirect" as opposed to "dereference" consistently
- add extra links
Fixes#3779.
R=r, rsc, iant, ken
CC=golang-dev
https://golang.org/cl/6326059
- receiving from a closed channel returns immediately
- in the ,ok form, the 2nd result is of type bool, not
just boolean (gc and ggcgo agree).
Per dsymonds' suggestion.
R=r, rsc, ken, iant, dsymonds
CC=golang-dev
https://golang.org/cl/6333057
They have been deleted from package unsafe.
Also delete their appearance in exp/types.
Fixes#3338.
R=golang-dev, dsymonds
CC=golang-dev
https://golang.org/cl/5847056
- Define what it means for two identifiers to be unique.
- The current spec is incorrect about exported
identifiers: for instance, it excluded fields
of non-exported types of exported variables
from being exported. It is easier to leave
the detailed specification away and let the
rest of the spec govern access of exported
identifiers.
- The current spec is incorrect about qualified
identifiers: It simply required that an identifier
be exported to be valid in a qualified identifier.
However, qualified identifiers can only access
exported identifiers declared in the package
block of the imported package.
Fixes#1551.
R=r, rsc, iant, ken
CC=golang-dev
https://golang.org/cl/5711043
- more idiomatic examples of pointer types
- show use of _ in examples of function types
- remove "legal:" qualification in examples
for consistency
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/5711054
When we first wrote the method set definition, we had long
discussions about whether method sets applied to all types
or just named types, and we (or at least I) concluded that it
didn't matter: the two were equivalent points of view, because
the only way to introduce a new method was to write a method
function, which requires a named receiver type.
However, the addition of embedded types changed this.
Embedding can introduce a method without writing an explicit
method function, as in:
var x struct {
sync.Mutex
}
var px *struct {
sync.Mutex
}
var _, _ sync.Locker = &x, px
The edits in this CL make clear that both &x and px satisfy
sync.Locker. Today, gccgo already works this way; 6g does not.
R=golang-dev, gri, iant, r
CC=golang-dev
https://golang.org/cl/5702062
Or, depending on your point of view, make the
comparisons satisfy any surrounding boolean type.
Also, fix a few foo_bar -> fooBar in code fragments.
Fixes#2561.
R=golang-dev, r, bradfitz, gri, iant, kevlar
CC=golang-dev
https://golang.org/cl/5671096
Drop reference to "machine type." Specify that integer
overflow must be an error. Drop requirement that exponent
must be 128 bits--that's a lot. Clarify that floating point
expressions may be rounded, including intermediate values.
This is a reworking of https://golang.org/cl/5577068/ .
Fixes#2789.
R=r, rsc, r, gri, ken, ken, iant
CC=golang-dev, remyoudompheng
https://golang.org/cl/5655049
The paragraph describing the ellipses presents the horizontal ellipsis
in two different fonts and at least on my screen they look too different.
Switch to a consistent rendering.
Of small steps are great journeys undertaken.
R=golang-dev, rsc, dsymonds, gri
CC=golang-dev
https://golang.org/cl/5650055
Other wordings are possible but defining this as one
of the ways to proceed means we don't have to add
language about this to the description of select.
Fixes#2825.
R=golang-dev, gri, r
CC=golang-dev
https://golang.org/cl/5643062
Document that indirection through a nil pointer will panic.
Explain function invocation.
This section will need more work, but it's a start.
Fixes#1865.
Fixes#2252.
R=rsc, iant, r
CC=golang-dev
https://golang.org/cl/5532114
This allows HTML pages to specify arbitrary data in a header:
<!--{
"Title": "The page title",
...
}-->
replacing the old style comments:
<!-- title The page title -->
R=gri, rsc, r, bradfitz, dsymonds
CC=golang-dev
https://golang.org/cl/5532093
- define "0-sized"
- add clarifying sentence to pointer comparison
- removed notion "location" which was used only in pointer comparisons
and which was never defined
Fixes#2620.
R=r, rsc, iant
CC=golang-dev
https://golang.org/cl/5528053
Several places mentioned tokens spanning "multiple lines"
which is not a well-defined term in the spec; newline is.
R=golang-dev, rsc, r
CC=golang-dev
https://golang.org/cl/5490046
This change guarantees that whether the line ending convention
when the source is created includes carriage returns is irrelevant
to the value of the string. See issue 680.
The compilers do not yet implement this.
R=golang-dev, adg, r, gri, rsc, iant
CC=golang-dev
https://golang.org/cl/5491043
Also, clarify when interface comparison panics and
that comparison to nil is a special syntax rather than
a general comparison rule.
R=r, gri, r, iant, cw, bradfitz
CC=golang-dev
https://golang.org/cl/5440117
This is a spec correction, not a language change.
The implementations have behaved like this for years
(and there are tests to that effect), and elsewhere in
the spec true and false are defined to be untyped
boolean constants.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/5477047
This has always been true, but we lost it from the spec
somewhere along the way, probably when we disallowed
the general 'pointer to anything sliceable' slice case.
R=gri
CC=golang-dev
https://golang.org/cl/5437121
The allowed conversions before and after are:
type Tstring string
type Tbyte []byte
type Trune []rune
string <-> string // ok
string <-> []byte // ok
string <-> []rune // ok
string <-> Tstring // ok
string <-> Tbyte // was illegal, now ok
string <-> Trune // was illegal, now ok
Tstring <-> string // ok
Tstring <-> []byte // ok
Tstring <-> []rune // ok
Tstring <-> Tstring // ok
Tstring <-> Tbyte // was illegal, now ok
Tstring <-> Trune // was illegal, now ok
Update spec, compiler, tests. Use in a few packages.
We agreed on this a few months ago but never implemented it.
Fixes#1707.
R=golang-dev, gri, r
CC=golang-dev
https://golang.org/cl/5421057
An experiment: allow structs to be copied even if they
contain unexported fields. This gives packages the
ability to return opaque values in their APIs, like reflect
does for reflect.Value but without the kludgy hacks reflect
resorts to.
In general, we trust programmers not to do silly things
like *x = *y on a package's struct pointers, just as we trust
programmers not to do unicode.Letter = unicode.Digit,
but packages that want a harder guarantee can introduce
an extra level of indirection, like in the changes to os.File
in this CL or by using an interface type.
All in one CL so that it can be rolled back more easily if
we decide this is a bad idea.
Originally discussed in March 2011.
https://groups.google.com/group/golang-dev/t/3f5d30938c7c45ef
R=golang-dev, adg, dvyukov, r, bradfitz, jan.mercl, gri
CC=golang-dev
https://golang.org/cl/5372095
This is true of the existing implementations, and I think
it is an important property to guarantee.
R=golang-dev, r, borman, r
CC=golang-dev
https://golang.org/cl/5321058
This is not a language change.
Added paragraphs specifying which conversions
yield results that are constants.
R=r, rsc, iant, ken
CC=golang-dev
https://golang.org/cl/4515176
This is (indirectly) a language change. Per e-mail discussion
on golang-dev.
Fixes#1943.
R=rsc, iant, r, ken
CC=golang-dev
https://golang.org/cl/4581058
- Added some additional examples.
- 6g appears to implement this semantics already.
Fixes#658.
R=rsc, r, iant, ken
CC=golang-dev
https://golang.org/cl/4538119
Note: This is not a spec change.
The spec was not clear on the result type of
constant shift expressions. Made it more
explicit and added additional examples.
Also: Remove paragraph on send expressions (they
are statements, now).
Fixes#1708.
R=rsc, r, iant, r
CC=golang-dev
https://golang.org/cl/4517074
This CL proposes some subtle language changes
in an attempt to clarify the semantics of range
clauses and simplify uses of maps.
- nil maps behave like empty maps; but attempting
to set a value in a nil map causes a run-time panic
- nil channels are never ready for communication;
sending or reading from a nil channel blocks forever
- if there is only one index iteration variable in a
range clause and len(range expression) would be a constant,
the range expression is not evaluated.
(was discrepancy with len/cap before)
- the notion of what is a constant expression len(x)
for (pointer to) arrays x has been generalized and
simplified (can still be syntactically decided)
(before: more restrictive syntactic rule that was not
consistently implemented)
Fixes#1713.
R=r, rsc, iant, ken2, r2, bradfitz, rog
CC=golang-dev
https://golang.org/cl/4444050
No language change.
- added a few examples with parentheses
- added a corresponding sentence to assignments
(this explicitly permits: (_) = 0, currently allowed by 6g,
gofmt, but marked as an error by gccgo).
R=rsc, r, iant
CC=golang-dev
https://golang.org/cl/4446071
The CL introduces inconsistencies with respect to
the use of parentheses/grouping of receive operations.
««« original CL description
spec: narrow syntax for expression and select statements
This is not a language change, it simply expresses the
accepted cases explicitly in the respective productions.
R=rsc, r, iant
CC=golang-dev
https://golang.org/cl/4428057
»»»
R=golang-dev, rsc1
CC=golang-dev
https://golang.org/cl/4444080
This is not a language change, it simply expresses the
accepted cases explicitly in the respective productions.
R=rsc, r, iant
CC=golang-dev
https://golang.org/cl/4428057
Computer people have an agglutinating streak that I like to resist.
As a time of execution: run time.
As an adjective: run-time.
As a noun: run-time support/code/library.
Signed,
Mr. Pedant.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/4252048
- added an example to Type declarations section clarifying the
situation brought up with issue 1324
- slightly re-ordered paragraphs in Types section
- added separate heading for method set section and refer to it
from elsewhere in the spec
- no language changes
R=rsc, r, iant, ken2, r2
CC=golang-dev
https://golang.org/cl/4145043
Removed most of the detailed examples about handing panics
from the go spec since it's now covered by Effective Go.
R=r, rsc, iant, ken2
CC=golang-dev
https://golang.org/cl/4128058
This change makes it legal to pass a struct value as receiver
outside the package declaring the struct even if the respective
struct type has non-exported fields.
This is a backwards-compatible language change motivated by the
fact that it is already possible to circumvent the assignment
restriction when calling methods through interfaces (see issue
1402).
R=r, rsc, iant, ken2
CC=golang-dev
https://golang.org/cl/3926044
Prior to this CL, there were two requirements about the
package name main.
1. The package that sits at the root of the import graph
(the one where program execution begins)
must be named main.
2. No other package in the program can be named main.
This CL only removes requirement #2, which can be done
without changing any other Go documentation.
The new wording and formatting is such that removing
requirement #1 can be done by deleting a single line,
but making that change is explicitly outside the scope
of this CL, because it would require changes to other
documentation at the same time.
R=gri, r, gri1
CC=golang-dev
https://golang.org/cl/4126053
These are syntactical changes to better reflect the communication
operator's new status in the language.
- sending to a channel is now done via a send statement
- there is no binary communication operation anymore which
leads to a reduction of the number of precedence levels
from 6 to 5 (yeah!)
- small semantic change: since a send operation is not part
of the expression syntax anymore, a <- send operator is
binding weaker than any other operator now
- receiving from a channel is done as before via the unary
receive expression
- communication clauses in select statement now can contain
send statements or receive expressions
R=rsc, r, iant, ken2, gri1
CC=golang-dev
https://golang.org/cl/3973051
Close of closed channel panics.
Receive from closed channel never panics,
even if done repeatedly.
Fixes#1349.
Fixes#1419.
R=gri, iant, ken2, r, gri1, r2, iant2, rog, albert.strasheim, niemeyer, ejsherry
CC=golang-dev
https://golang.org/cl/3989042
The default float type is not very useful but for the most basic applications.
For instance, as it is now, using the math package requires conversions for float
variables (the arguments for math functions are usually float64). Typical real
applications tend to specify the floating point precision required.
This proposal removes the predeclared types float and complex. Variable declarations
without type specification but with constant floating point or complex initializer
expressions will assume the type float64 or complex128 respectively.
The predeclared function cmplx is renamed to complex.
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/3423041
- use math.Sqrt instead of Math.sqrt
- use float64 for Point fields to match math.Sqrt
- distinguish between Point and Point3D for clarity
- add alignment sizes for complex types
R=r, rsc, iant, ken2
CC=golang-dev
https://golang.org/cl/3420041
Added a "return" to the end of an example which previously threw a compile error if used.
R=golang-dev, gri
CC=golang-dev
https://golang.org/cl/3052041
Consistency argument: A valid Go program should
remain valid after stripping leading and trailing
whitespace. This was not true so far if the last
text in the source was a line comment.
R=iant, ken2, r, rsc, r2
CC=golang-dev
https://golang.org/cl/2906041
For elements which are themselves composite literals, the type may
be omitted if it is identical to the element type of the containing
composite literal.
R=r, rsc, iant, ken2
CC=golang-dev
https://golang.org/cl/2661041
Background: The current spec is imprecise with respect to the parsing ambiguity
for composite literals: It says that the ambiguity arises when the TypeName form
of the LiteralType is used. The following code:
if (B) {} ...
is not using the TypeName form (but the parenthesized TypeName form) and thus
could be interpreted as:
if ((B){}) ...
instead of
if B {} ...
Both compilers and gofmt choose the latter interpretation. One could fix the
spec by making the clause regarding the parsing ambiguity more precise ("...using
the _possibly parenthesized_ TypeName form of the LiteralType..."). The alternative
(chosen here) is to simply disallow parenthesized literal types. Except for a single
test case (test/parentype.go) there appears to be no Go code under $GOROOT containing
parenthesized literal types. Furthermore, parentheses are never needed around a
literal type for correct parsing.
R=golang-dev
CC=golang-dev
https://golang.org/cl/1913041
We didn't mention this explicitly during our discussions,
but I think it fits the "identical types are spelled identically"
rule that we used.
R=gri, iant, ken2, r, rsc1
CC=golang-dev
https://golang.org/cl/1698043
Specifically:
- introduced notion of "underlying type"
- removed notion of type compatibility
- consolidated rules about assignment compatibility in
assignment compatibility section
- be consistent with specyfing that nil is the value
for uninitialized variables that can be nil (this
was not specified clearly for pointers, functions, interfaces)
- added/fixed various related links throughout
- clarify language on conversions
R=rsc, r, iant, ken2
CC=golang-dev
https://golang.org/cl/1536041
Specifically:
- Simplified definition of comparison compatibility and folded into
section on comparison operators since it's only used there.
This is a small language change/cleanup. As a consequence:
- An interface value may now be compared against a non-interface value.
- Channels with opposite directions cannot be compared directly anymore
(per discussion with rsc).
R=rsc, r, iant, ken2
CC=golang-dev
https://golang.org/cl/1462041
and adjust conversion rules.
Also:
- clarification of type identity (no language change)
- adjust special rules for channel assignment/comparison to
require identical element types (in correspondence to non-
recursiveness of type compatibility)
R=rsc, iant, ken2, r
CC=golang-dev
https://golang.org/cl/1376042
- also added links to the respective sections
- reformatted section for better (source html) readability
- added missing closing </li>'s in many places
R=r
CC=golang-dev
https://golang.org/cl/1220041
Also: Actual map key must be assignment-compatible with
formal map key type.
Fixes#357.
R=r, iant, rsc, ken2
CC=golang-dev
https://golang.org/cl/673042
This was discussed with Ken and Russ. 6g has been changed,
and both compilers now work this way. This avoids a double
type conversion--first to the float type, then to the complex
type.
R=r
CC=golang-dev, ken2, rsc
https://golang.org/cl/419042
