Two tests start a goroutine which runs exportSend, and then
the tests run importReceive. exportSend creates an export
channel. importReceive asks to receive values on that
channel. Because exportSend runs in a separate goroutine,
it's possible for the export client to receive the request for
values on the channel, from importReceive, before the
goroutine actually creates the export channel. That causes an
error: "export: no such channel: exportedSend". This patch
avoids the race by creating the export channel before starting
the goroutine.
There does not seem to be a similar race condition in the
tests which send data in the other direction.
R=r
CC=golang-dev
https://golang.org/cl/2026045
- refine/define Scope, Object, and Type structures
(note: scope.go has the addition of types, the rest is only re-organized
for better readability)
- implemented top-level of type checker:
resolve global type declarations (deal with double decls, cycles, etc.)
- temporary hooks for checking of const/var declarations, function/method bodies
- test harness for fine-grained testing (exact error locations)
with initial set of tests
This is a subset of the code for easier review.
R=rsc
CC=golang-dev
https://golang.org/cl/1967049
Based on the observation that a great number of the types that
are copied or compared in interfaces, maps, and channels are
word-sized, this uses specialized copy and equality functions
for them that use a word instead of 4 or 8 bytes. Seems to yield
0-6% improvements in performance in the benchmarks I've run.
For example, with the regexp benchmarks:
Before:
regexp.BenchmarkLiteral 500000 3.26 µs/op
regexp.BenchmarkNotLiteral 100000 13.67 µs/op
regexp.BenchmarkMatchClass 100000 18.72 µs/op
regexp.BenchmarkMatchClass_InRange 100000 20.04 µs/op
regexp.BenchmarkReplaceAll 100000 27.85 µs/op
After:
regexp.BenchmarkLiteral 500000 3.11 µs/op
regexp.BenchmarkNotLiteral 200000 13.29 µs/op
regexp.BenchmarkMatchClass 100000 17.65 µs/op
regexp.BenchmarkMatchClass_InRange 100000 18.49 µs/op
regexp.BenchmarkReplaceAll 100000 26.34 µs/op
R=rsc
CC=golang-dev
https://golang.org/cl/1967047
This way, if you later want to check things in,
you can (with appropriate authorization).
Using plain http leads to the cryptic error
abort: HTTP Method Not Allowed.
R=r
CC=golang-dev
https://golang.org/cl/1983050
Commands written in Go depend on Go packages, so they
cannot be built by src/cmd/make.bash. They have been
built by src/make.bash after all the packages are done, but
we want to be able to use cgo (written in Go) during the build
of package net. To make this possible, build the commands
from src/pkg/Makefile instead of src/make.bash, so that they
are included in the package dependency analysis.
R=r
CC=golang-dev
https://golang.org/cl/1972046
Due to header key normalization/typo, the server never responds with
the protocol header in place. This breaks all (draft76) applications
that are using the protocol-header.
R=ukai, rsc
CC=golang-dev
https://golang.org/cl/1969046
Despite the name, URL escaping is for a small subpiece of the URL only.
This particular URL is being emitted in an <a href=""> tag and in that
context it should be HTML escaped, not URL escaped.
In addition to fixing a bug, this change cuts a dependency chain
from go/doc to net, clearing the way for use of cgo
(which imports go/doc) during the compilation of package net.
R=gri
CC=golang-dev
https://golang.org/cl/1960050
(Assumed to be in $PATH. all.bash ensures that
during the main build and the user must ensure it
when running commands like gotest or gomake
by hand. This belonged in the earlier CL but I missed it.)
R=r
CC=golang-dev
https://golang.org/cl/1967048
The runtime only passes 32 bits of file offset,
but the kernel wants 64 bits, so have to add
zeros explicitly in a copy of the arguments.
R=adg, Martin Neubauer
CC=golang-dev
https://golang.org/cl/1933044
Cannot use paren field in Node because all
instances of a given symbol name use the same Node.
Fixes#1022.
R=ken2
CC=golang-dev
https://golang.org/cl/2015043
Augmented ASTs may contain cycles. Keep
track of already printed objects and refer
to them with a line number instead of
printing them again.
R=rsc
CC=golang-dev
https://golang.org/cl/1998042
go/ast: implement Fprint and print functions to
print AST nodes
gofmt: print AST nodes by setting -ast flag
R=rsc, r
CC=golang-dev
https://golang.org/cl/1981044
Function to create a GoString with a known length so it can contain NUL
bytes anywhere in the string. Some C libraries have strings like this.
R=rsc
CC=golang-dev
https://golang.org/cl/2007042
(Argument: For any *PtrValue p, it should
always be possible to do: p.PointTo(p.Elem()),
even if p.Elem() is nil.)
Fixes#1028.
R=rsc
CC=golang-dev, r
https://golang.org/cl/1938044
tabwriter: Introduce a new flag StripEscape to control
if tabwriter.Escape chars should be stripped or passed
through unchanged.
go/printer: Don't modify tabwriter.Escape'd text. This
involved a new implementation of the internal trimmer
object.
Does not affect formatting of any existing code under
$GOROOT/src and $GOROOT/misc.
Fixes#1030.
R=rsc
CC=golang-dev
https://golang.org/cl/1943045
the solution must work around a weakness in the reflection library:
there is no way to do type-safe conversions under reflection.
R=rsc
CC=golang-dev
https://golang.org/cl/2000041
This changeset implements client certificate support in crypto/tls
for both handshake_server.go and handshake_client.go
The updated server implementation sends an empty CertificateAuthorities
field in the CertificateRequest, thus allowing clients to send any
certificates they wish. Likewise, the client code will only respond
with its certificate when the server requests a certificate with this
field empty.
R=agl, rsc, agl1
CC=golang-dev
https://golang.org/cl/1975042
- change ast.Ident back to contain the name and adjust all dependent code
- identifier object information will be added again through an optional
typechecker phase (in the works).
- remove tracking of scopes in parser - it's easier to do this in a separate
phase (in the works)
- in godoc, generate popup info table directly instead of through a formatter
for simpler data flow (at the expense of a little bit more code)
Runs all tests.
As a result of this change, the currently shown popup information
(const, var, type, func, followed by identifier name) will not be
shown anymore temporarily.
R=rsc
CC=golang-dev
https://golang.org/cl/1994041
draft-ietf-6man-text-addr-representation-07 will introduce
a canonical textual representation format for IPv6 address.
R=rsc
CC=golang-dev
https://golang.org/cl/1856047
Also, if the header is bad, exit with a non-zero status.
Other calls to Brdline in the tree, by category:
Reading symbol name from object file:
./cmd/5l/obj.c:486: name = Brdline(f, '\0');
./cmd/6l/obj.c:535: name = Brdline(f, '\0');
./cmd/8l/obj.c:564: name = Brdline(f, '\0');
./libmach/sym.c:292: cp = Brdline(bp, '\0');
Reading archive header line (fixed, short):
./cmd/gc/lex.c:287: if((a = Brdline(b, '\n')) == nil)
./cmd/gc/lex.c:303: if((p = Brdline(b, '\n')) == nil)
Reading object file header line (fixed, short):
./cmd/ld/lib.c:421: line = Brdline(f, '\n');
Reading undefined symbol list (unused code):
./cmd/ld/lib.c:773: while((l = Brdline(b, '\n')) != nil){
Implementing Brdstr:
./libbio/brdstr.c:36: p = Brdline(bp, delim);
The symbol names ones will cause a problem loudly if they
fail: they'll error out with symbol name too long. This means
that you can't define an enormous struct without giving the
type a name and then stick it in an interface, because the
type's symbol name will be too long for the object file.
Since this will be a loud failure instead of a silent one,
I'm willing to wait until it comes up in practice.
R=r
CC=golang-dev
https://golang.org/cl/1982041
The previous set was spotty, incomplete, and confusing.
This CL proposes a regular, clean set with clearer names.
It's also complete. Many existing methods will be deprecated,
but not in this CL. Ditto for the tests.
R=rsc, gri
CC=golang-dev, rog
https://golang.org/cl/1946041
go/scanner: return information on semicolon (real or inserted) when
found in source
go/parser: better error message when a semicolon is found unexpectedly
For instance, if an unexpected semicolon is found that was automatically
inserted, the parser error message is now:
"expected '}', found newline"
Fixes#1006.
R=rsc
CC=golang-dev
https://golang.org/cl/1936044
Fixes#971.
Parse/ParseFile methods of Template now match template.Parse and .ParseFile methods.
Also made tests being run on Parse and ParseFile be run on Template.ParseFile as well.
R=r
CC=golang-dev
https://golang.org/cl/1741059
Rect types.
The actual image representation is unchanged. A future change will
replace the {[][]color} with {[]color, stride int, r Rectangle} and
possibly a clip region.
The draw.Color, draw.Point and draw.Rect types will be removed in a
future change. Trying to do it in this one polluted the diff with
trivia.
R=r, rsc
CC=golang-dev
https://golang.org/cl/1918047
Fixed:
* SetString calls in bitwise tests
* Aliasing problem with self bitwise test
* One test case that was just flat out wrong
* Slice panics in nat.or and nat.xor
* Aliasing problems in Int.And, Int.AndNot, Int.Or, Int.Xor
Fixes#1007.
R=gri
CC=golang-dev
https://golang.org/cl/1895049
Generic text-based network protcol library for SMTP-like protocols.
HTTP and NNTP should be changed to use this package,
and I expect that SMTP and POP3 will be able to use it too.
R=cemeyer, nigeltao_golang, r
CC=golang-dev, petar-m
https://golang.org/cl/889041
Basically these functions are implemented the same way as the
corresponding functions in the strings package. Test functions
are implemented for IndexRune and FieldsFunc.
Additionally two typos are fixed in packages bytes and strings.
R=r
CC=golang-dev
https://golang.org/cl/1696062
Returns R14 and R15 to the available register pool.
Plays more nicely with ELF ABI C code.
In particular, our signal handlers will no longer crash
when a signal arrives during execution of a cgo C call.
Fixes#720.
R=ken2, r
CC=golang-dev
https://golang.org/cl/1847051
- don't allow parenthesized receiver base types or anonymous fields
- fixed a couple of other omissions
- adjusted gofmt test script
- removed several TODOs
R=rsc
CC=golang-dev
https://golang.org/cl/1897043
Besides being more correct, it protects against people accidentally
exchanging the permission and open mode arguments to Open.
R=rsc
CC=golang-dev
https://golang.org/cl/1904045
RFC 5280, 4.2.1.2 says:
SubjectKeyIdentifier ::= KeyIdentifier
KeyIdentifier ::= OCTET STRING
Previously, we were failing to unwrap the second level of OCTET STRING
encoding.
Fixes#993.
R=rsc
CC=golang-dev
https://golang.org/cl/1917044
Fixes#980.
Made it return an empty array, rather than crash.
Added relevant test cases to strings.
R=golang-dev, r
CC=golang-dev
https://golang.org/cl/1914041
Goroutine 1:
Call Read on read half of pipe, entering pipeHalf.rw.
Check ioclosed field, which is false.
Send data to p.c1
Wait for response on p.c2.
Goroutine 2:
Call Close on read half of pipe, entering pipeHalf.close.
Set closed field.
Send error to p.cclose.
Set ioclosed field.
Send 1 to p.done.
Return and exit goroutine.
Goroutine 3:
This is the goroutine running pipe.run, and for some reason
it has started late.
Read error from p.rclose; set rerr and continue.
Read 1 from p.done; increment ndone and continue.
Read data from r1 (sent by goroutine 1); set r1 = nil and continue
Now goroutine 1 is waiting for a response, and goroutine 3 is
waiting for something else to happen.
This patch fixes the race by having the runner check whether
the read half is closed when it is asked for read data, and
similarly for the corresponding race on the write half.
This patch also fixes the similar race in which ndone gets
bumped up to 2 while there is a reader or writer waiting.
There is still another race to fix. It is possible for the
read half and the write half to both be closed, and for the
runner goroutine to exit, all before the runner goroutine sees
the request from a reader. E.g., in the above, have goroutine
2 also close the write half, and have goroutine 3 see both
done messages before it sees the request from goroutine 1.
R=rsc
CC=golang-dev
https://golang.org/cl/1862045
For the Windows version of syscall Errstr, set the
FORMAT_MESSAGE_IGNORE_INSERTS value of the FormatMessage
Flags argument when there are no values to insert.
R=rsc, brainman
CC=golang-dev
https://golang.org/cl/1868043
SNI (Server Name Indication) is a way for a TLS client to
indicate to the server which name it knows the server by. This
allows the server to have several names and return the correct
certificate for each (virtual hosting).
PeerCertificates returns the list of certificates presented by
server.
R=r
CC=golang-dev
https://golang.org/cl/1741053
OCSP is the preferred X.509 revocation mechanism. X.509 certificates
can contain a URL from which can be fetched a signed response saying
"this certificate is valid until $x" (where $x is usually 7 days in the
future). These are called OCSP responses and they can also be included
in the TLS handshake itself ("OCSP stapling")
R=rsc, r
CC=golang-dev
https://golang.org/cl/1875043
- don't lose empty lines after labels
- canonicalize number of line breaks
- gofmt src misc, fixes a couple of irregular breaks
R=rsc
CC=golang-dev
https://golang.org/cl/1843044
but with less precision than hardware counterparts.
fixed a number of tests to output BUG when they failed.
changed the runner to distinghuish between output
and output containing ^BUG
R=rsc
CC=dho, golang-dev
https://golang.org/cl/1778041
strings.ToTitle converts all characters to title case, which for consistency with the
other To* functions it should continue to do. This CL adds string.Title, which
does a proper title-casing of the string.
A similar function for package bytes will follow once this is settled.
Fixes#933.
R=rsc
CC=golang-dev
https://golang.org/cl/1869042
Add support for ASN.1 ENUMERATED types.
Add a magic type, asn1.Flag, for the cases where the presence of an
empty explicit tag is semantically meaningful.
Add support for GeneralizedTime.
R=rsc
CC=golang-dev
https://golang.org/cl/1684055
out of floating constant multiply
2. added rounding code to "const fix=float"
to allow up to 29 (Mpscale) bits of
slop and still get an exact fixed constant.
fixes#931
R=rsc
CC=golang-dev
https://golang.org/cl/1692055
//line 10 units.y
which is equiv to c
#line 10 units.y
the purpose is to generate diagnostics
that correctly point to preprocessed source.
R=rsc
CC=golang-dev
https://golang.org/cl/1863042
Somewhat of a work-in-progress (in that MIME is a large spec), but this is
functional and enough for discussion and/or code review.
In addition to the unit tests, I've tested with curl and Chrome with
a variety of test files, making sure the digests of files are unaltered
when read via a multipart Part.
R=rsc, adg, dsymonds1, agl1
CC=golang-dev
https://golang.org/cl/1681049
* remember #defined names, so that C.stdout can refer
to the real name (on OS X) __stdoutp.
* better handling of #defined constant expressions
* allow n, err = C.strtol("asdf", 0, 123) to get errno as os.Error
* write all output files to current directory
* don't require gcc output if there was no input
Fixes#533.
Fixes#709.
Fixes#756.
R=r
CC=dho, golang-dev, iant
https://golang.org/cl/1734047
Also in this CL:
* Removed util.go, as its functionality is in big
* Removed some semicolons from the code generated by gen.go
* Added a generate target to Makefile
* Removed an outdated TODO from value.go
R=gri
CC=golang-dev
https://golang.org/cl/1780042
Use io/ioutil.TempFile with default os.TempDir for temporary test files.
For os_test.go temporary test files, use a local file system and OS
independent directory names. Avoid problems with NFS.
Fixes#848.
R=adg
CC=golang-dev
https://golang.org/cl/1806043
- sign to determine if a value is < 0, == 0, > 0
- abs to compute absolute value
- Rat.IsInt to test if a rational number is representable as an integer
R=rsc
CC=golang-dev
https://golang.org/cl/1761042
* Allow an exponent part. This is necessary for exp/eval.
* Fix a bug for input that had no numbers after the decimal.
* In Int.SetString, allow a leading + sign.
* In Int.SetString, error if the input is "-" with no number.
* In nat.scan, normalize the resulting nat.
R=gri
CC=golang-dev
https://golang.org/cl/1733045
updated thread.c to provide destroylock, which seems to be
required to link.
updated README with different virtualization programs.
R=golang-dev, adg
CC=golang-dev
https://golang.org/cl/1746047
(Here, quoted strings are the official AMD names.)
The amd64 "movsxd" instruction, when invoked
with a 64-bit REX prefix, moves and sign extends
a 32-bit value from register or memory into a
64-bit register. 6.out.h spells this MOVLQSX.
6.out.h also includes MOVLQZX, the zero extending
version, which it implements as "movsxd" without
the REX prefix. Without the REX prefix it's only sign
extending 32 bits to 32 bits (i.e., not doing anything
to the bits) and then storing in a 32-bit register.
Any write to a 32-bit register zeros the top half of the
corresponding 64-bit register, giving the advertised effect.
This particular implementation of the functionality is
non-standard, because an ordinary 32-bit "mov" would
do the same thing.
Because it is non-standard, it is often mishandled or
not handled by binary translation tools like valgrind.
Switching to the standard "mov" makes the binaries
work better with those tools.
It's probably useful in 6c and 6g to have an explicit
instruction, though, so that the intent of the size
change is clear. Thus we leave the concept of MOVLQZX
and just implement it by the standard "mov" instead of
the non-standard 32-bit "movsxd".
Fixes#896.
R=ken2
CC=golang-dev
https://golang.org/cl/1733046
With these changes, goinstall is now able to use branches
maintained with Bazaar located in Launchpad.
Project aliases such as /project and /project/series are
supported in addition to specific user or team branches
such as /~user/project/branch. Temporary branches under
the +junk special project are also supported.
As a curious side effect, since Launchpad is able to import
code from other locations, they can be indirectly
accessible too if desired.
R=rsc
CC=golang-dev
https://golang.org/cl/1699050
The Makefile and cgo now rewrite / to _ when creating the path.
The .so for gosqlite.googlecode.com/hg/sqlite is named
cgo_gosqlite.googlecode.com_hg_sqlite.so, and then 6l and 8l
both include a default rpath of $GOROOT/pkg/$GOOS_$GOARCH.
This should make it easier to move binaries from one system
to another.
Fixes#857.
R=iant, r
CC=golang-dev
https://golang.org/cl/1700048
This is the Replace I suggested in the review of CL 1114041.
It's true that we already have
regexp.MustCompile(regexp.QuoteMeta(old)).ReplaceAll(s, new)
but because this Replace is doing a simpler job it is
simpler to call and inherently more efficient.
I will add the bytes implementation and tests to the
CL after the strings one has been reviewed.
R=r, cw
CC=golang-dev
https://golang.org/cl/1731048
For generating non-self-signed certs we need to be able to specify a
public key (for the signee) which is different from the private key (of
the signer).
R=rsc
CC=golang-dev
https://golang.org/cl/1741045
One goroutine started up and was waiting in rw. Then another
goroutine decided to close the pipe. The closing goroutine
stalled calling p.io.Lock() in pipeHalf.close. (This happened
in gccgo). If the closing goroutine had been able to set the
ioclosed flag, it would have gone on to tell the runner that
the pipe was closed, which would then send an EINVAL to the
goroutine sleeping in rw. Unlocking p.io before sleeping in
rw avoids the race.
R=rsc, rsc1
CC=golang-dev
https://golang.org/cl/1682048
Revision: 5885c9d10f created syscall_bsd.go for code used
by Darwin and other *BSDs, which should have included
FreeBSD. mksyscall.sh to generate new zsyscall_freebsd_386.go.
Fixes#862.
R=golang-dev, adg
CC=golang-dev
https://golang.org/cl/1701048
Ping IDs should be limited to 16-bits. Fix failure printing.
R=rsc
CC=golang-dev, jean-christophe smith <jeanchristophe.smith
https://golang.org/cl/1682043
Currently to install a command, you have to manually
goinstall each of the remote packages that it depends on.
This patch lets goinstall P work where P is
contains files in package main.
It does not actually build the package, but
it installs all of its dependencies and prints a message
to that effect.
R=rsc
CC=golang-dev
https://golang.org/cl/1301043
x.go:13: cannot use t (type T) as type Reader in assignment:
T does not implement Reader (Read method requires pointer receiver)
x.go:19: cannot use q (type Q) as type Reader in assignment:
Q does not implement Reader (missing Read method)
have read()
want Read()
x.go:22: cannot use z (type int) as type Reader in assignment:
int does not implement Reader (missing Read method)
x.go:24: too many arguments to conversion to complex: complex(1, 3)
R=ken2
CC=golang-dev
https://golang.org/cl/1736041
This CL replaces my earlier https://golang.org/cl/1640044/show
in which Continue handling was explicit. Instead, this CL makes
it automatic. Reading from Body() is an implicit acknowledgement
that the request headers were fine and the body is wanted. In that
case, the 100 Continue response is written automatically when the
request continues the "Expect: 100-continue" header.
R=rsc, adg
CC=golang-dev
https://golang.org/cl/1610042
While we're at it, clean up and test the code to guarantee we see every byte when
the text is erroneous UTF-8.
Fixes#866.
R=rsc
CC=golang-dev
https://golang.org/cl/1712042
This shortens, simplifies and regularizes the code significantly.
(Improvements to reflect could make another step.)
Passes all.bash.
One semantic change occurs: The String() method changes
behavior. It used to run only for string formats such as %s and %q.
Instead, it now runs whenever the item has the method and the
result is then processed by the format as a string. Besides the
regularization, this has three effects:
1) width is honored for String() items
2) %x works for String() items
3) implementations of String that merely recur will recur forever
Regarding point 3, example from the updated documentation:
type X int
func (x X) String() string { return Sprintf("%d", x) }
should cast the value before recurring:
func (x X) String() string { return Sprintf("%d", int(x)) }
R=rsc
CC=golang-dev
https://golang.org/cl/1613045
Change TrimRight and TrimLeft to use these functions.
Incidentally fix minor bug in TrimRight.
Add some test cases for this.
YMMV whether it's worth saving the closure allocation.
R=r, r2
CC=golang-dev, hoisie, rsc
https://golang.org/cl/1198044
Cannot assume that g == m->curg at time of signal.
Must save actual g and restore.
Fixes flaky crashes with messages like
throw: malloc mlookup
throw: malloc/free - deadlock
throw: unwindstack on self
throw: free mlookup
(and probably others) when running cgo.
R=iant
CC=golang-dev
https://golang.org/cl/1648043
There's only one Go object file per package now,
so there's no need to parse the metadata and merge
metadata from multiple files. Just save the original
and use it as __.PKGDEF verbatim.
R=r
CC=golang-dev
https://golang.org/cl/1647042
This avoids a crash when using cgo where glibc's malloc thinks
that it can use some of the memory following the symbol table.
This fails because the symbol table is mapped read-only, which
affects the whole page.
R=rsc
CC=golang-dev
https://golang.org/cl/1616042
Although technically incorrect, we want this in order to parse X.509
certificates where a wildcard hostname ("*.example.com") has been put
into a PrintableString.
Fixes#850.
R=rsc
CC=golang-dev
https://golang.org/cl/1615043
* Code for assignment, conversions now mirrors spec.
* Changed some snprint -> smprint.
* Renamed runtime functions to separate
interface conversions from type assertions:
convT2I, assertI2T, etc.
* Correct checking of \U sequences.
Fixes#840.
Fixes#830.
Fixes#778.
R=ken2
CC=golang-dev
https://golang.org/cl/1303042
(caused certain files to not be idempotent under gofmt)
- corrected golden files for go/printer
- slightly simplified some code in nodes.go (no impact on formatting)
- these changes have no impact on gofmt output of .go files under src, misc
fallthrough statement considered harmful!
R=rsc
CC=golang-dev
https://golang.org/cl/1593042
Time (us) to draw a 200x200 src onto an image.RGBA dst with the Over
operator. Each case was measured three separate times, and the sorted
times are reported.
Fill case (where src is an image.ColorImage):
Before:
7438, 7511, 7526
After:
3073, 3087, 3102
Copy case (where src is an image.RGBA):
Before:
9506, 9510, 9563
After:
5262, 5300, 5344
R=rsc
CC=golang-dev
https://golang.org/cl/1532041
Also fix an interface bug: white-space-delimited doesn't work well for cases like "%d, %d" on "23, 23")
R=rsc
CC=golang-dev
https://golang.org/cl/1502041
charclass to avoid unnecessarily iterating over ranges.
Also, use the fact that IntVector is an []int to avoid method calls.
On my machine, this brings us from ~27500 ns/op to ~17500 ns/op in the benchmark I've added (it is also faster in the case where a range check
doesn't help, added a benchmark for this too.)
I'd also like to propose that "[]", and "[^]" be disallowed. They aren't useful as far as I can tell, they aren't widely supported, and they make reasoning about character classes a bit more complicated.
R=r
CC=golang-dev
https://golang.org/cl/1495041
On error, ReadSlice is defined to return the available data
and advance past that data, but it was not behaving that
way for err == ErrBufferFull, making it harder for callers to
handle well.
R=r
CC=golang-dev
https://golang.org/cl/1480041
- implement scanning for all renamed types
(compiler bug stops complex from being renamable,
so it can't be tested but the code is there)
- %q %x for strings
- error handling now done with panic/recover
R=rsc
CC=golang-dev
https://golang.org/cl/1458041
- implement scanning for all renamed types
(compiler bug stops complex from being renamable,
so it can't be tested but the code is there)
- %q %x for strings
- error handling now done with panic/recover
R=rsc
CC=golang-dev
https://golang.org/cl/1458041
For example, earlier, the regexp would accept SIGQUEUE_MAX
as a unix signal with name SIGQUEUE. Now it is ignored.
R=iant
CC=golang-dev, golang-nuts, joel.sherrill
https://golang.org/cl/1452041
Also, move Do() to vector.go, so that Do() for IntVector and StringVector
is autogenerated.
The only files edited are Makefile, defs.go, and vector.go. The rest are
autogenerated with "make generate".
R=r
CC=golang-dev, hoisie
https://golang.org/cl/1435041
- reimplement integer scanning to handle renamed basic integer types
- start implementation of Fscanf and Scanf; not yet ready for general use.
This intermediate CL is a useful checkpoint. A similar change needs to be
done for float and complex, but it seemed worth getting the approach
reviewed before making those changes.
R=rsc
CC=golang-dev
https://golang.org/cl/1418041
It turns out that _cgo_malloc is used, via cmalloc in
runtime/cgocall.c, which is called by code generated by out.go
for the ·_C_CString function. I can't find a call to
_cgo_free, but given _cgo_malloc we might as well keep
_cgo_free. This patch fixes it so that it should work on
amd64.
R=rsc
CC=golang-dev
https://golang.org/cl/1399041
Fix Rectangle.Clip.
It could return a non-canonical rectangle if its arguments
did not overlap.
e.g. Rect(0, 0, 10, 10).Clip(Rect(0, 20, 10, 30)) -> Rect(0, 20, 10, 10)
R=rsc, r
CC=golang-dev, golang-dev
https://golang.org/cl/1290041
ModInverse is just a small wrapper around GcdInt, but it's nice to
have in order to be clear about what one is doing in other code.
R=gri, agl1
CC=golang-dev
https://golang.org/cl/1244045
Still to do:
- composite types
- user-defined scanners
- format-driven scanning
The package comment will be updated when more of the functionality is in place.
R=rsc
CC=golang-dev
https://golang.org/cl/1252045
On my laptop, time to prepare and write 800x600 pixels over the
socket falls from 125-ish ms to 80-ish ms.
Thanks to Roger Peppe for the suggestion.
R=r
CC=golang-dev
https://golang.org/cl/1228044
cannot allocate an audomatic temp
while real registers are allocated.
there is a chance that the automatic
will be allocated to one of the
allocated registers. the fix is to
not registerize such variables.
R=rsc
CC=golang-dev
https://golang.org/cl/1202042
Time to draw.Draw a 200x200 image fell from 18.4ms (and 1 malloc) to
5.6ms (and 0 mallocs). It's still relatively slow since it assumes
nothing about the src or mask images, but it does remove the malloc.
There are existing faster, more specialized paths for copies, fills
and image glyph masks.
Also added a "compare to a slow but obviously correct implementation"
check to draw_test.go.
R=rsc, r
CC=golang-dev
https://golang.org/cl/1223044
- implemented setWord, use it where setUint64 is wrong
- divLarge: use fast mulWW, divWW; implemented mulWW, divWW
- better assembly code for addMulVVW
R=rsc
CC=golang-dev
https://golang.org/cl/1258042
Previously the compiler would just emit "internal compiler error" when
trying to compute the remainder of floats or complex types.
R=rsc
CC=golang-dev
https://golang.org/cl/1243041
- support for binary prefix 0b (to match fmt.Format)
- renamed nat.new -> nat.setUint64 for consistency
- more tests
R=r
CC=golang-dev
https://golang.org/cl/1233041
Also update range of Phase and Polar due to signed zero.
[Phase(cmplx(-1, +0)) = pi and Phase(cmplx(-1, -0)) = -pi]
R=rsc, r
CC=golang-dev
https://golang.org/cl/1235041
Timings (as for change 1122043) go from 49ms to 48ms ish. It's
mostly lost in the noise, but it probably doesn't hurt.
R=r
CC=golang-dev
https://golang.org/cl/1179041
To draw.Draw a 32x32 image.Alpha 10000 times,
Before: 633ms with 10000 mallocs
After: 49ms with 0 mallocs
These times are just blitting an image.Alpha, and do not include
rasterizing a glyph's vector contours to an image.Alpha.
The "generic" test case in draw_test.go tests this fast path.
R=rsc
CC=golang-dev
https://golang.org/cl/1122043
- removed last argument (n) from all core arithmetic routines;
instead, use the length of the result
- simplified nat.make implementation and chose a better capacity
for new values, removed a TODO in the process
Changing the constant e from 1 (old) to 4 (new) improved
pidigits -s -n 10000 by ~9% (on a 3.06GHz Intel Core 2 Duo):
user 0m3.882s (old)
user 0m3.549s (new)
R=rsc
CC=golang-dev
https://golang.org/cl/1133043
This permits cgo callbacks to work when run in init code.
Otherwise cgocallback switches to the wrong stack address.
R=rsc
CC=golang-dev
https://golang.org/cl/1123043
- no need to make copies in cases of aliases
- removed deprecated internal shift functions
- minor unrelated simplifications
This change improves pidigits -s -n10000 by almost 20%:
user 0m6.156s (old)
user 0m4.999s (new)
(pidigits -s -n20000 goes from ~25s to ~19s)
R=rsc
CC=golang-dev
https://golang.org/cl/1149041
Because maps are mostly a hidden type, they must be
implemented using reflection values and will not be as
efficient as arrays and slices.
R=rsc
CC=golang-dev
https://golang.org/cl/1127041
- renamed Len -> BitLen, simplified implementation
- renamed old Div, Mod, DivMod -> Que, Rem, QuoRem
- implemented Div, Mod, DivMod (Euclidian definition, more
useful in a mathematical context)
- fixed a bug in Exp (-0 was possible)
- added extra tests to check normalized results everywhere
- uniformly set Int.neg flag at the end of computations
- minor cosmetic cleanups
- ran all tests
R=rsc
CC=golang-dev
https://golang.org/cl/1091041
Import _mulv from Inferno again, change R9 to R2.
Not sure what the other differences were for, but
they weren't working.
TBR=kaib
CC=golang-dev
https://golang.org/cl/1079041
built. Note that they are only built if a //export comment
appears, so other targets should only use them if there is
such a comment.
Fixes#723.
R=rsc
CC=golang-dev
https://golang.org/cl/1073041
When trying to regenerate src/pkg/runtime/darwin/386/defs.h
on a 64 bit capable Snow Leopard (OS X 10.6.3) system I
needed to add -f -m32 to godefs, as this OS and hardware
combination defaults to 64 bit compilation.
For safety, make the same change to the 32 bit FreeBSD
instructions in .../freebsd/defs.c. (Tested OK and no
problems introduced.)
R=golang-dev, rsc
CC=golang-dev
https://golang.org/cl/1052042
- fixed a couple of bugs in the process
(shift right was incorrect for negative numbers)
- added more tests and made some tests more robust
- changed pidigits back to using shifts to multiply
by 2 instead of add
This improves pidigit -s -n 10000 by approx. 5%:
user 0m6.496s (old)
user 0m6.156s (new)
R=rsc
CC=golang-dev
https://golang.org/cl/963044
This results in an improvement of > 35% for the existing Mul benchmark
using the same karatsuba threshold, and an improvement of > 50% with
a slightly higher threshold (32 instead of 30):
big.BenchmarkMul 500 6731846 ns/op (old alg.)
big.BenchmarkMul 500 4351122 ns/op (new alg.)
big.BenchmarkMul 500 3133782 ns/op (new alg., new theshold)
Also:
- tweaked calibrate.go, use same benchmark as for Mul benchmark
R=rsc
CC=golang-dev
https://golang.org/cl/1037041
Plus:
- calibration "test" - include in tests with gotest -calibrate
- basic Mul benchmark
- extra multiplication tests
- various cleanups
This change improves multiplication speed of numbers >= 30 words
in length (current threshold; found empirically with calibrate):
The multiplication benchmark (multiplication of a variety of long numbers)
improves by ~35%, individual multiplies can be significantly faster.
gotest -benchmarks=Mul
big.BenchmarkMul 500 6829290 ns/op (w/ Karatsuba)
big.BenchmarkMul 100 10600760 ns/op
There's no impact on pidigits for -n=10000 or -n=20000
because the operands are are too small.
R=rsc
CC=golang-dev
https://golang.org/cl/1004042
Rather than confuse people by linking to a spec we
don't fully support, let them just use the package docs.
R=r, rsc
CC=golang-dev
https://golang.org/cl/1016041