siddontang / go Goto Github PK

run following command and test case is not passed

go test ./... -test.v -test.cpu=1,2,4
go test ./... -test.v -test.short -test.race

siddontang，在查看你的时间轮代码时候，觉得(TimingWheel)After()函数中index计算不正确，为了验证猜想，我写了如下测试用例：

• https://github.com/AlexStocks/go-practice/blob/master/time/siddontang_time_wheel.go

输出是：expect time cost 100s, while real time cost is 110 s.
显然，当运行次数越多，你的timingwheel的时间误差将越多，我给你修正后的(TimingWheel)After()函数代码如下：
func (w *TimingWheel) After(timeout time.Duration) <-chan struct{} {
if timeout >= w.maxTimeout {
panic("timeout too much, over maxtimeout")
}

index := int(timeout / w.interval)
if 0 < index {
    index--
}
w.Lock()

index = (w.pos + index) % len(w.cs)

b := w.cs[index]

w.Unlock()

return b

}

再次运行我的测试用例后，输出如下：
expect time cost 100s, while real time cost is 100 s.

This output is perfect.

你好，我是不是要在run方法中w.onTicker()这行上面开一个协程去处理时间轮当前位置里面要处理的逻辑啊，例如关闭socket连接等操作

cb568a3 does not pass unit tests with Go 1.10. At least:

Testing: "/builddir/build/BUILD/go-cb568a3e5cc06256f91a2da5a87455f717eb33f4/_build/src/github.com/siddontang/go/filelock"
+ GOPATH=/builddir/build/BUILD/go-cb568a3e5cc06256f91a2da5a87455f717eb33f4/_build:/usr/share/gocode
+ go test -buildmode pie -compiler gc -ldflags '-extldflags '\''-Wl,-z,relro  '\'''
# github.com/siddontang/go/filelock
./file_lock_test.go:43: Fatal call has possible formatting directive %s
FAIL    github.com/siddontang/go/filelock [build failed]
+ GOPATH=/builddir/build/BUILD/go-cb568a3e5cc06256f91a2da5a87455f717eb33f4/_build:/usr/share/gocode
+ go test -buildmode pie -compiler gc -ldflags '-extldflags '\''-Wl,-z,relro  '\'''
# github.com/siddontang/go/list2
./list_test.go:60: Errorf format %p has arg p of wrong type int
./list_test.go:73: Errorf format %p has arg n of wrong type int
FAIL    github.com/siddontang/go/list2 [build failed]

siddontang，在查看你的时间轮代码时候，觉得(TimingWheel)After()函数中index计算不正确，为了验证猜想，我写了如下测试用例：

• https://github.com/AlexStocks/go-practice/blob/master/time/siddontang_time_wheel.go

输出是：expect time cost 100s, while real time cost is 110 s.
显然，当运行次数越多，你的timingwheel的时间误差将越多，我给你修正后的(TimingWheel)After()函数代码如下：
`func (w *TimingWheel) After(timeout time.Duration) <-chan struct{} {
if timeout >= w.maxTimeout {
panic("timeout too much, over maxtimeout")
}

index := int(timeout / w.interval)
if 0 < index {
    index--
}
w.Lock()

index = (w.pos + index) % len(w.cs)

b := w.cs[index]

w.Unlock()

return b

}`

再次运行我的测试用例后，输出如下：
expect time cost 100s, while real time cost is 100 s.

This output is perfect.

Description

go test failed bson_test

Reproduction guide

1. clone repo
2. run go test ./...

Observed behaviour

`----------------------------------------------------------------------
FAIL: bson_test.go:597: S.TestMarshalErrorItems

bson_test.go:600:
c.Assert(err, ErrorMatches, item.data)
... error string = "Attempted to marshal Raw kind 10 as a document"
... regex string = "Attempted to unmarshal Raw kind 10 as a document"

FAIL: bson_test.go:1429: S.TestObjectIdJSONUnmarshalingError

bson_test.go:1432:
c.Assert(err, ErrorMatches, Invalid ObjectId in JSON: "4d88e15b60f486e428412dc9A")
... error string = "invalid ObjectId in JSON: "4d88e15b60f486e428412dc9A""
... regex string = "Invalid ObjectId in JSON: "4d88e15b60f486e428412dc9A""`

Expected behaviour:

test passed.

System Info

OS: gnu/linux
golang: go version go1.9.2 linux/amd64