Justin Nuß [Tue, 26 Jan 2016 20:14:35 +0000 (21:14 +0100)]
strconv: Avoid allocation in AppendQuote*
The current implementations of the AppendQuote functions use quoteWith
(through Quote) for quoting the given value and appends the returned
string to the dst byte slice. quoteWith internally creates a byte slice
on each call which gets converted to a string in Quote.
This means the AppendQuote functions always allocates a new byte slice
and a string only to append them to an existing byte slice. In the case
of (Append)QuoteRune the string passed to quoteWith will also needs to
be allocated from a rune first.
Split quoteWith into two functions (quoteWith and appendQuotedWith) and
replace the call to Quote inside AppendQuote with appendQuotedWith,
which appends directly to the byte slice passed to AppendQuote and also
avoids the []byte->string conversion.
Also introduce the 2 functions quoteRuneWith and appendQuotedRuneWith
that work the same way as quoteWith and appendQuotedWith, but take a
single rune instead of a string, to avoid allocating a new string when
appending a single rune, and use them in (Append)QuoteRune.
Also update the ToASCII and ToGraphic variants to use the new functions.
Martin Möhrmann [Sat, 20 Feb 2016 00:51:32 +0000 (01:51 +0100)]
fmt: change padding functions to avoid package init
Move the decision if zero padding is allowed to doPrintf
where the other formatting decisions are made.
Removes some dead code for negative f.wid that was never used
due to f.wid always being positive and f.minus deciding if left
or right padding should be used.
New padding code writes directly into the buffer and is as fast
as the old version but avoids the cost of needing package init.
name old time/op new time/op delta
SprintfPadding-2 246ns ± 5% 245ns ± 4% ~ (p=0.345 n=50+47)
Change-Id: I7dfddbac8e328f4ef0cdee8fafc0d06c784b2711
Reviewed-on: https://go-review.googlesource.com/19957
Run-TryBot: Rob Pike <r@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Rob Pike <r@golang.org>
Matthew Dempsky [Fri, 26 Feb 2016 10:10:29 +0000 (02:10 -0800)]
cmd/compile: simplify lexinit and lexfini
Split the syms array into separate basicTypes and builtinFuncs arrays.
Also, in lexfini, instead of duplicating the code from lexinit to
declare the builtin identifiers in the user package, just import them
from builtinpkg like how importdot works.
Change-Id: Ic3b3b454627a46f7bd5f290d0e31443e659d431f
Reviewed-on: https://go-review.googlesource.com/19936
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Robert Griesemer <gri@golang.org>
Martin Möhrmann [Wed, 24 Feb 2016 22:22:40 +0000 (23:22 +0100)]
fmt: use public io.RuneScanner interface for ScanState reader
All io.Reader that are passed to newScanState in all the standard
library tests that implement io.RuneReader also implement io.RuneScanner.
Do not check on each call ScanState's UnreadRune that the used RuneReader
also implements the UnreadRune method by using a private interface.
Instead require the used Reader to implement the public RuneScanner
interface.
The extra implementation logic for UnreadRune is removed from ScanState.
Instead the readRune wrapper is extended to implement UnreadRune for the
RuneScanner interface. If the Reader passed to newScanstate does not
implement RuneScanner the readRune wrapper is used to implement the
missing functionality.
Note that a RuneReader that does not implement RuneScanner will also
be wrapped by runeRead which was not the case before.
Performance with the readRune wrapper is better than without before.
Add benchmark to compare performance with and without using the
readRune wrapper.
cmd/compile: recognize more memory runs in generated algs
The old implementation assumed that all memory runs
were terminated by non-memory fields.
This isn't necessarily so.
They might be terminated by padding or blank fields.
For example, given
type T struct {
a int64
b byte
c, d, e int64
}
the old implementation did a memory comparison on a+b, on c, and on d+e.
Instead, check for memory runs at the beginning of every round.
This now generates a memory comparison on a+b and on c+d+e.
Also, delete some now-dead code.
Change-Id: I66bffb111420adf6919bd708e4fb3a1e1f07fadd
Reviewed-on: https://go-review.googlesource.com/19841 Reviewed-by: David Crawshaw <crawshaw@golang.org>
Robert Griesemer [Fri, 26 Feb 2016 21:32:28 +0000 (13:32 -0800)]
cmd/compile: track pragmas in lexer rather than global variables
By using a Pragma bit set (8 bits) rather than 8 booleans, also
reduce Func type size by 8 bytes (208B -> 200B on 64bit platforms,
116B -> 108B on 32bit platforms).
Change-Id: Ibb7e1f8c418a0b5bc6ff813cbdde7bc6f0013b5a
Reviewed-on: https://go-review.googlesource.com/19966 Reviewed-by: Dave Cheney <dave@cheney.net>
Matthew Dempsky [Fri, 26 Feb 2016 06:10:48 +0000 (22:10 -0800)]
cmd/compile: fix contrived line number errors
If a general comment contains multiple newline characters, we can't
simply unread one and then re-lex it via the general whitespace lexing
phase, because then we'll reset lineno to the line before the "*/"
marker, rather than keeping it where we found the "/*" marker.
Also, for processing imports, call importfile before advancing the
lexer with p.next(), so that lineno reflects the line where we found
the import path, and not the token afterwards.
Fixes #14520.
Change-Id: I785a2d83d632280113d4b757de0d57c88ba2caf4
Reviewed-on: https://go-review.googlesource.com/19934 Reviewed-by: Robert Griesemer <gri@golang.org>
Run-TryBot: Matthew Dempsky <mdempsky@google.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Ian Lance Taylor [Thu, 25 Feb 2016 18:35:19 +0000 (10:35 -0800)]
cmd/compile: change Func.{Dcl,Inldcl} from NodeList to slice
A slice uses less memory than a NodeList, and has better memory locality
when walking the list.
This uncovered a tricky case involving closures: the escape analysis
pass when run on a closure was appending to the Dcl list of the OCLOSURE
rather than the ODCLFUNC. This happened to work because they shared the
same NodeList. Fixed with a change to addrescapes, and a check to
Tempname to catch any recurrences.
This removes the last use of the listsort function outside of tests.
I'll send a separate CL to remove it.
Unfortunately, while this passes all tests, it does not pass toolstash
-cmp. The problem is that cmpstackvarlt does not fully determine the
sort order, and the change from listsort to sort.Sort, while generally
desirable, produces a different ordering. I could stage this by first
making cmpstackvarlt fully determined, but no matter what toolstash -cmp
is going to break at some point.
Matthew Dempsky [Thu, 25 Feb 2016 22:58:03 +0000 (14:58 -0800)]
cmd: stop looking for __.(GO)?SYMDEF entries in archives
The Go toolchain stopped creating them before Go 1.3, so no point in
worrying about them today.
History:
- Git commit 250a091 added cmd/ar, which wrote Plan 9 __.SYMDEF
entries into archive files.
- golang.org/cl/6500117 renamed __.SYMDEF to __.GOSYMDEF. (Notably,
the commit message suggests users need to use Go nm to read symbols,
but even back then the toolchain did nothing with __.(GO)?SYMDEF files
except skip over them.)
- golang.org/cl/42880043 added the -pack flag to cmd/gc to directly
produce archives by the Go compiler, and did not write __.GOSYMDEF
entries.
- golang.org/cl/52310044 rewrote cmd/pack in Go, and removed support
for producing __.GOSYMDEF entries.
Richard Miller [Wed, 27 Jan 2016 16:18:34 +0000 (16:18 +0000)]
sync/atomic: new file for plan9_arm support
Atomic load/store/add/swap routines, as for other ARM platforms, but with DMB inserted
for load/store (assuming that "atomic" also implies acquire/release memory ordering).
Michael McConville [Sun, 3 Jan 2016 01:50:57 +0000 (20:50 -0500)]
crypto/rand: use the getentropy syscall on OpenBSD
Go already supports Linux's getrandom, which is a slightly modified
version of getentropy.
getentropy was added in OpenBSD 5.6. All supported versions of OpenBSD
include it so, unlike with Linux and getrandom, we don't need to test
for its presence.
Jeff R. Allen [Sun, 16 Aug 2015 12:59:23 +0000 (14:59 +0200)]
misc/cgo: fix wrong quote character
Single quotes to not expand variables inside of them.
Change-Id: I4a0622c0aebfc1c3f9d299f93f7a8253893b5858
Reviewed-on: https://go-review.googlesource.com/13661 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Dmitry Vyukov [Fri, 26 Feb 2016 15:14:31 +0000 (16:14 +0100)]
sync/atomic: reduce test in short mode
In normal mode the test runs for 9+ seconds on my machine (48 cores).
But the real problem is race mode, in race mode it hits 10m test timeout.
Reduce test size in short mode. Now it runs for 100ms without race.
Rick Arnold [Fri, 26 Feb 2016 03:24:07 +0000 (22:24 -0500)]
net/url: support query string without values
Previously, RawQuery was used to indicate the presence of a query
string in url.URL. However, this approach was not able to differentiate
between URLs that have no query string at all (http://foo.bar/) and
those that have a query with no values (http://foo.bar/?).
Add a ForceQuery field to indicate the latter form of URL and use it
in URL.String to create a matching URL with a trailing '?'.
Dmitry Vyukov [Fri, 26 Feb 2016 09:43:09 +0000 (10:43 +0100)]
runtime: unwire g/m in dropg always
Currently dropg does not unwire locked g/m.
This is unnecessary distiction between locked and non-locked g/m.
We always restart goroutines with execute which re-wires g/m.
First, this produces false sense that this distinction is necessary.
Second, it can confuse some sanity and cross checks. For example,
if we check that g/m are unwired before we wire them in execute,
the check will fail for locked g/m. I've hit this while doing some
race detector changes, When we deschedule a goroutine and run
scheduler code, m.curg is generally nil, but not for locked ms.
Austin Clements [Fri, 26 Feb 2016 02:01:16 +0000 (21:01 -0500)]
runtime: remove workbuf logging
Early in Go 1.5 we had bugs with ownership of workbufs, so we added a
system for tracing their ownership to help debug these issues.
However, this system has both CPU and space overhead even when
disabled, it clutters up the workbuf API, the higher level gcWork
abstraction makes it very difficult to mess up the ownership of
workbufs in practice, and the tracing hasn't been enabled or needed
since 5b66e5d nine months ago. Hence, remove it.
Benchmarks show the usual noise from changes at this level, but little
overall movement.
name old time/op new time/op delta
XBenchGarbage-12 2.48ms ± 1% 2.47ms ± 0% -0.68% (p=0.000 n=21+21)
Datong Sun [Sat, 20 Feb 2016 08:39:07 +0000 (02:39 -0600)]
crypto/x509: better documentation for ParsePKIXPublicKey
The existing documentation for ParsePKIXPublicKey is difficult to understand
and the return type of the parsed public key are not mentioned explicitly.
Descriptions about types of public key supported, as well as an example on
how to use type assertions to determine return type of a parsed public key
has been added.
Matthew Dempsky [Fri, 26 Feb 2016 02:17:31 +0000 (18:17 -0800)]
cmd/compile: simplify error sorting
Errors have unique seq values (their index within the errors slice),
so errcmp never needs to fallback to sorting by message text.
Moreover, comparing by original index is exactly the purpose of using
a stable sort algorithm (and sort.Stable was added in Go 1.2), so we
really only need to compare by lineno.
Matthew Dempsky [Fri, 26 Feb 2016 00:07:04 +0000 (16:07 -0800)]
cmd/compile: rationalize (lex)?lineno handling
Previously, many error messages inconsistantly used either lexlineno
and lineno. In general this works out okay because they're almost
always the same. The only exceptional case is after lexing a
multi-line raw string literal, where lineno will be the line number of
the opening quote and lexlineno is the line number of the closing
quote.
This CL makes the compiler's error message more consistent:
- Lexer error messages related to invalid byte sequences (i.e., NUL
bytes, bad UTF-8 sequences, and non-initial BOMs) are emitted at
lexlineno (i.e., the source line that contains the invalid byte
sequence).
- All other error messages (notably the parser's "syntax errors") now
use lineno. The minor change from this is that bogus input like:
package `
bogus`
will emit "syntax error: unexpected string literal, expecting name"
error at line 1, instead of line 2.
- Instead of maintaining prevlineno all the time, just record it
when/where actually needed and not already available elsewhere (which
turns out to be just one function).
- Lastly, we remove the legacy "syntax error near ..." fallback in
Yerror, now that the parser always emits more detailed syntax error
messages.
Change-Id: Iaf5f784223d0385fa3a5b09ef2b2ad447feab02f
Reviewed-on: https://go-review.googlesource.com/19925 Reviewed-by: Robert Griesemer <gri@golang.org>
Austin Clements [Thu, 18 Feb 2016 14:38:49 +0000 (09:38 -0500)]
runtime: eliminate unused _Genqueue state
_Genqueue and _Gscanenqueue were introduced as part of the GC quiesce
code. The quiesce code was removed by 197aa9e, but these states and
some associated code stuck around. Remove them.
Austin Clements [Mon, 23 Nov 2015 23:45:18 +0000 (18:45 -0500)]
runtime: use only per-P gcWork
Currently most uses of gcWork use the per-P gcWork, but there are two
places that still use a stack-based gcWork. Simplify things by making
these instead use the per-P gcWork.
Austin Clements [Mon, 23 Nov 2015 23:44:03 +0000 (18:44 -0500)]
runtime: pass gcWork to markroot
Currently markroot uses a gcWork on the stack and disposes of it
immediately after marking one root. This used to be necessary because
markroot was called from the depths of parfor, but now that we call it
directly and have ready access to a gcWork at the call site, pass the
gcWork in, use it directly in markroot, and share it across calls to
markroot from the same P.
Austin Clements [Tue, 9 Feb 2016 23:37:41 +0000 (18:37 -0500)]
runtime: remove noescape hacks from gcWork
When gcWork was first introduced, the compiler's escape analysis
wasn't good enough to detect that that method receiver didn't escape,
so we had to hack around this.
Now that the compiler can figure out this for itself, remove these
hacks.
Austin Clements [Tue, 9 Feb 2016 23:30:38 +0000 (18:30 -0500)]
runtime: remove unnecessary clears of the heap bitmap
Currently we clear the heap bitmap of a span both when we allocate
that span *and* when we free it. There's no point in doing both, and
we definitely have to write the heap bitmap when we allocate a span
for pointer-sized objects, so switch to clearing only when we allocate
a span.
This results in a slight overall performance improvement; however,
most of the benchmarks that get slower are very short, while the
longer benchmarks generally got faster.
name old time/op new time/op delta
XBenchGarbage-12 2.48ms ± 1% 2.47ms ± 1% -0.58% (p=0.000 n=91+91)
Austin Clements [Mon, 15 Feb 2016 21:50:12 +0000 (16:50 -0500)]
runtime: document non-obvious requirement on sudog.elem
The channel code must not allow stack splits between when it assigns a
potential stack pointer to sudog.elem (or sudog.selectdone) and when
it makes the sudog visible to copystack by putting it on the g.waiting
list. We do get this right everywhere, but add a comment about this
subtlety for future eyes.
Austin Clements [Tue, 9 Feb 2016 22:24:26 +0000 (17:24 -0500)]
runtime: fix heapBitsSweepSpan comment
Currently the heapBitsSweepSpan comment claims that heapBitsSweepSpan
sets the heap bitmap for the first two words to dead. In fact, it sets
the first *four* words to scalar/dead. This is important because first
two words don't actually have a dead bit, so for objects larger than
two words it *must* set a dead bit in third word to reset the object
to a "noscan" state. For example, we use this in heapBits.hasPointers
to detect that an object larger than two words is noscan.
Dmitriy Dudkin [Thu, 25 Feb 2016 19:48:57 +0000 (21:48 +0200)]
cmd/go: clear cmd cache to avoid duplicate loads errors
go get -u all command updates all packages including standard
commands. We need to get commands evicted from their cache to
avoid loading old versions of the packages evicted from the
packages cache.
Ian Lance Taylor [Thu, 25 Feb 2016 19:43:45 +0000 (11:43 -0800)]
doc/go1.7: mention CallersFrames and Frames
Change-Id: I73ae6a6837a6dcf75b3b8f431d97a18348e01a42
Reviewed-on: https://go-review.googlesource.com/19921 Reviewed-by: Ian Lance Taylor <iant@golang.org>
David Crawshaw [Thu, 18 Feb 2016 11:31:57 +0000 (06:31 -0500)]
cmd/compile: remove rtype.ptrToThis
Simplifies some code as ptrToThis was unreliable under dynamic
linking. Now the same type lookup is used regardless of execution
mode.
A synthetic relocation, R_USETYPE, is introduced to make sure the
linker includes *T on use of T, if *T is carrying methods.
Changes the heap dump format. Anything reading the format needs to
look at the last bool of a type of an interface value to determine
if the type should be the pointer-to type.
Reduces binary size of cmd/go by 0.2%.
For #6853.
Change-Id: I79fcb19a97402bdb0193f3c7f6d94ddf061ee7b2
Reviewed-on: https://go-review.googlesource.com/19695 Reviewed-by: Keith Randall <khr@golang.org>
Run-TryBot: David Crawshaw <crawshaw@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Mikio Hara [Fri, 19 Feb 2016 08:39:27 +0000 (17:39 +0900)]
net: re-enable TestDualStack{TCP,UDP}Listener on dragonfly
It looks like the latest DragonFly BSD kernels, at least 4.4 and above,
have finished working on handling of shared IP control blocks. Let's
re-enbale test cases referring to IP control blocks and see what
happens.
Updates #13146.
Change-Id: Icbe2250e788f6a445a648541272c99b598c3013d
Reviewed-on: https://go-review.googlesource.com/19406 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Mikio Hara [Wed, 24 Feb 2016 02:59:49 +0000 (11:59 +0900)]
net: make TestGoLookupIPWithResolverConfig robust
It crashes when the node under the test is shaken up.
-- FAIL: TestGoLookupIPWithResolverConfig (11.73s)
panic: interface conversion: error is nil, not *net.DNSError [recovered]
panic: interface conversion: error is nil, not *net.DNSError
goroutine 23 [running]:
panic(0x2e2620, 0xc820181440)
/go/src/runtime/panic.go:483 +0x3f3
testing.tRunner.func1(0xc820136d80)
/go/src/testing/testing.go:467 +0x192
panic(0x2e2620, 0xc820181440)
/go/src/runtime/panic.go:441 +0x4f6
net.TestGoLookupIPWithResolverConfig(0xc820136d80)
/go/src/net/dnsclient_unix_test.go:358 +0x7ca
testing.tRunner(0xc820136d80, 0x49ddc0)
/go/src/testing/testing.go:473 +0x98
created by testing.RunTests
/go/src/testing/testing.go:582 +0x892
exit status 2
Change-Id: I9631f41a3c73f3269c7e30d679c025ae64d71a98
Reviewed-on: https://go-review.googlesource.com/19870 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Mikio Hara [Thu, 25 Feb 2016 00:55:40 +0000 (09:55 +0900)]
net: fix typo
Change-Id: Ic828256efe0f50a3e11a25d85092d7531b342d2e
Reviewed-on: https://go-review.googlesource.com/19873 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Robert Griesemer [Thu, 25 Feb 2016 00:17:49 +0000 (16:17 -0800)]
cmd/compile: adjust starting token value
The actual values assigned to tokens was inherited from the yacc-based
grammar. With the most recent cleanups, all single-char tokens such as
commas, semis, parens, etc., that get returned from lexer.next simply
as their Unicode values are below utf8.RuneSelf (i.e., 7bit ASCII).
Lower the initial starting value for named token constants accordingly.
Change-Id: I7eb8e584dbb3bc7f9dab849d1b68a91320cffebd
Reviewed-on: https://go-review.googlesource.com/19913 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Tal Shprecher [Sat, 20 Feb 2016 00:02:54 +0000 (16:02 -0800)]
cmd/asm: fix EOF message on operand parsing errors.
If the parsing of an operand completes but the parser thinks there
is more to read, return an "expected end of operand" error message
instead of "expected EOF." This also removes extra "asm: " prefixes
in error strings since "asm: " is already set as the global log
prefix.
Fixes #14071
Change-Id: I7d621c1aea529a0eca3bcba032359bd25b3e1080
Reviewed-on: https://go-review.googlesource.com/19731 Reviewed-by: Rob Pike <r@golang.org>
Keith Randall [Wed, 24 Feb 2016 17:12:51 +0000 (09:12 -0800)]
cmd/compile: Drop references to Prog structs after each function
Don't accumulate a massive list of Prog structs during
compilation and write them all out at the end of compilation.
Instead, convert them to code+relocs (or data+relocs) after each
function is compiled.
Track down a few other places that were keeping Progs alive
and nil them out so the Progs get GCd promptly.
Saves ~20% in peak memory usage for the compiler. Surprisingly not much
help speed-wise (only because we end up doing more GCs. With a
compensating GOGC=120, it does help a bit), but this provides a base for
more changes (e.g. reusing a cache of Progs).
Change-Id: I838e01017c228995a687a8110d0cd67bf8596407
Reviewed-on: https://go-review.googlesource.com/19867
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Josh Bleecher Snyder <josharian@gmail.com> Reviewed-by: Ian Lance Taylor <iant@golang.org>
Robert Griesemer [Wed, 24 Feb 2016 06:04:20 +0000 (22:04 -0800)]
cmd/compile: factor our literal lexing from main lexer function
Further reduces complexity of lexer.next which is now readable.
Also removes the need to initialize various local variables in
each next call even if they are not used for the current token.
No measurable performance change for `time go build -a net/http`
(best of 5 runs): difference < 0.3% (in the noise).
Jure Ham [Tue, 23 Feb 2016 10:41:27 +0000 (11:41 +0100)]
sort: fix for nondeterministic less function in quicksort pivot
Fixes #14377
Change-Id: I130a6e1b8bc827db44efd0a74e759b894ecc4977
Reviewed-on: https://go-review.googlesource.com/19823 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
kortschak [Thu, 18 Feb 2016 10:49:03 +0000 (21:19 +1030)]
cmd/go, go/build: add support for Fortran
This change adds support for Fortran files (.f, .F, .for, .f90) to the
go tool, in a similar fashion to Objective-C/C++. Only gfortran is
supported out of the box so far but leaves other Fortran compiler
toolchains the ability to pass the correct link options via CGO_LDFLAGS.
A simple test (misc/cgo/fortran) has been added and plugged into the
general test infrastructure. This test is only enabled when the $FC
environment variable is defined (or if 'gfortran' was found in $PATH.)
Derived from CL 4114.
Change-Id: Ifc855091942f95c6e9b17d91c17ceb4eee376408
Reviewed-on: https://go-review.googlesource.com/19670 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Run-TryBot: Ian Lance Taylor <iant@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Catalin Nicutar [Sun, 21 Feb 2016 18:32:25 +0000 (18:32 +0000)]
cmd/vet: add a check for tests with malformed names
According to golang.org/pkg/testing the first character after Test has
to be non-lowercase. Functions that don't conform to this are not
considered tests and are not loaded which can cause surprises.
This change adds a check to warn about Test-like functions in a _test
file that are not actually run by go test.
Moved over from https://go-review.googlesource.com/#/c/19466/
Change-Id: I2f89676058b27a0e35f721bdabc9fa8a9d34430d
Reviewed-on: https://go-review.googlesource.com/19724 Reviewed-by: Ian Lance Taylor <iant@golang.org> Reviewed-by: Rob Pike <r@golang.org>
Robert Griesemer [Tue, 23 Feb 2016 07:07:30 +0000 (23:07 -0800)]
cmd/compile: towards simpler and faster lexing: always use getr
Always reading runes (rather than bytes) has negligible overhead
(a simple if at the moment - it can be eliminated eventually) but
simplifies the lexer logic and opens up the door for speedups.
In the process remove many int conversions that are now not needed
anymore.
Also, because identifiers are now more easily recognized, remove
talph label and move identifier lexing "in place".
Also, instead of accepting all chars < 0x80 and then check for
"frogs", only permit valid characters in the first place. Removes
an extra call for common simple tokens and leads to simpler logic.
`time go build -a net/http` (best of 5 runs) seems 1% faster.
Assuming this is in the noise, there is no noticeable performance
degradation with this change.
Change-Id: I3454c9bf8b91808188cf7a5f559341749da9a1eb
Reviewed-on: https://go-review.googlesource.com/19847 Reviewed-by: Matthew Dempsky <mdempsky@google.com>
Run-TryBot: Robert Griesemer <gri@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Mikio Hara [Tue, 23 Feb 2016 09:06:31 +0000 (18:06 +0900)]
net: rename test files
This change renames {ipraw,tcp,udp,unix}_test.go to
{ipraw,tcp,udp,unix}sock_test.go for clarification. Also moves
NSS-related system configuration test helpers into main_conf_test.go and
main_noconf_test.go.
Change-Id: I28ba1e8ceda7b182ee3aa85f0ca3321388ba45e2
Reviewed-on: https://go-review.googlesource.com/19787 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Shenghou Ma [Wed, 24 Feb 2016 00:20:41 +0000 (19:20 -0500)]
cmd/compile/internal/gc: update comment after c2go
Change-Id: I02c60f6c767e917a8ed3772c2773fe266f781e44
Reviewed-on: https://go-review.googlesource.com/19834 Reviewed-by: Ian Lance Taylor <iant@golang.org>
Matthew Dempsky [Tue, 23 Feb 2016 23:29:19 +0000 (15:29 -0800)]
cmd/compile: remove parser lineno hack for issue #13267
After golang.org/cl/19652 removed the bizarre lexlineno{++,--}
statements for parsing canned imports, this hack for #13267 is no
longer necessary:
$ echo -n 0 > /tmp/0.go
$ go tool compile /tmp/0.go
/tmp/0.go:1: syntax error: package statement must be first
Apparently setting lexlineno to 2 while parsing the canned imports
caused prevlineno and lineno to also be set to 2. After we finished
parsing imports and restored lexlineno to 1, since "package" is the
first token in a source file, we'll have fixed lineno = 1, but
prevlineno was still set to 2.
Change-Id: Ibcc49fe3402264819b9abb53505631f7a0ad4a36
Reviewed-on: https://go-review.googlesource.com/19859 Reviewed-by: Robert Griesemer <gri@golang.org>
Keith Randall [Tue, 23 Feb 2016 18:54:36 +0000 (10:54 -0800)]
cmd/compile: keep JMPs around with -N
When -N, make sure we don't drop every instruction from
a block, even ones which would otherwise be empty.
Helps keep line numbers around for debugging, particularly
for break and continue statements (which often compile
down to nothing).
Fixes #14379
Change-Id: I33722c4f0dcd502f146fa48af262ba3a477c959a
Reviewed-on: https://go-review.googlesource.com/19854
Run-TryBot: Keith Randall <khr@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org> Reviewed-by: Minux Ma <minux@golang.org>
Prashant Varanasi [Sat, 20 Feb 2016 17:43:15 +0000 (09:43 -0800)]
net: fix for DialTimeout errors with large timeout
The existing implementation converts the deadline time to an int64,
but does not handle overflow. If the calculated deadline is negative
but the user specified deadline is in the future, then we can assume
the calculation overflowed, and set the deadline to math.MaxInt64.
Fixes #14431
Change-Id: I54dbb4f02bc7ffb9cae8cf62e4e967e9c6541ec6
Reviewed-on: https://go-review.googlesource.com/19758 Reviewed-by: Ian Lance Taylor <iant@golang.org> Reviewed-by: Mikio Hara <mikioh.mikioh@gmail.com>
Robert Griesemer [Tue, 23 Feb 2016 19:15:56 +0000 (11:15 -0800)]
text/tabwriter: clarify documentation
More clearly distinguish between tab-terminated cells
which are part of an (aligned) column, and non-tab terminated
cells which are not part of a column. Added additional examples.
For #14412.
Change-Id: If72607385752e221eaa2518238b11f48fbcb8a90
Reviewed-on: https://go-review.googlesource.com/19855 Reviewed-by: Alan Donovan <adonovan@google.com>