Indeed, it was about precision, the issue is with Windows, not with Go.
I've used only one time.Now call and the planets have aligned.
package main
import (
"time"
"math"
"fmt"
)
func main() {
now := time.Now()
datetime := now.Add(time.Hour * 24 * 7 * 4 * 12 * 3)
seconds := -1 * int(now.Sub(datetime).Seconds())
a := 29030400
x := float64(seconds)/float64(a)
fmt.Println("input:", x, "floor:", math.Floor(x))
}
Result on both systems: input: 3 floor: 3.
Thank you all!
On Saturday, May 5, 2018 at 7:54:27 AM UTC+3, Andrei Avram wrote:
>
> Yes, this could be it. First I was fooled by println rounding 2.9 to 3,
> thinking floor is not working properly on Linux.
>
> Thanks!
>
> On Saturday, May 5, 2018 at 1:49:34 AM UTC+3, Steven Hartland wrote:
>>
>> You could be seeing a side effect on fact that windows only ticks every
>> 15ms.
>>
>> On Fri, 4 May 2018 at 21:56, Andrei Avram <[email protected]> wrote:
>>
>>> This code is extracted from something real. Someone on the team noticed
>>> the unit tests (which I wrote on a Linux machine) were failing on their
>>> Windows machine.
>>> I'll continue studying this and come back if I find something new.
>>>
>>> Thank you both for your time, I appreciate it!
>>>
>>>
>>> On Friday, May 4, 2018 at 11:42:00 PM UTC+3, speter wrote:
>>>
>>>> To file a bug (and have it treated seriously) you would need to
>>>> demonstrate that it is causing problems in a program that actually does
>>>> something useful, not just a pathological code sample. My expectation
>>>> would
>>>> be that once you start doing some real processing, the difference between
>>>> time.Now() invocations becomes non-zero, that is this "issue" doesn't
>>>> reproduce with "real" practical programs.
>>>>
>>>>
>>>> On Fri, May 4, 2018 at 10:26 PM, Andrei Avram <[email protected]>
>>>> wrote:
>>>>
>>>>> But I don't think it is because Windows is so much faster than Linux
>>>>>>
>>>>>
>>>>> Yeah... :) I was exploring all cases.
>>>>>
>>>>>
>>>>> and/or the way the time package implemented is different between
>>>>>> Windows and Linux
>>>>>
>>>>>
>>>>> Could this be considered a possible Go bug and would it worth opening
>>>>> an issue on Github?
>>>>>
>>>>>
>>>>> On Friday, May 4, 2018 at 10:57:46 PM UTC+3, speter wrote:
>>>>>>
>>>>>> So on Linux it's working as expected. In the playground time is
>>>>>> "virtual"; the start time is fixed and it doesn't progress unless you
>>>>>> explicitly Sleep -- because you don't Sleep in your program, time.Now()
>>>>>> returns the same time on both invocation.
>>>>>>
>>>>>> So the only slightly surprising part is that on Windows, time as
>>>>>> measured by time.Now() doesn't progress between the two statements. But
>>>>>> I
>>>>>> don't think it is because Windows is so much faster than Linux. :) I'd
>>>>>> guess this is because the way Windows manages time is different from how
>>>>>> Linux manages it, and/or the way the time package implemented is
>>>>>> different
>>>>>> between Windows and Linux. The result of the cross-platform differences
>>>>>> seems to be less precision on Windows -- at least in this specific
>>>>>> scenario.
>>>>>>
>>>>>> Peter
>>>>>>
>>>>>>
>>>>>> On Fri, May 4, 2018 at 9:33 PM, Andrei Avram <[email protected]>
>>>>>> wrote:
>>>>>>
>>>>>>> Peter and Ian, you are both wright regarding the printing. Still, on
>>>>>>> Linux I have different values than on Go Playground.
>>>>>>>
>>>>>>> Regarding the speed, Peter, do you think that on Windows the two
>>>>>>> calls just run faster every time?
>>>>>>>
>>>>>>> On Friday, May 4, 2018 at 9:07:15 PM UTC+3, speter wrote:
>>>>>>>>
>>>>>>>> b is slightly less than 3 because there is a bit of time between
>>>>>>>> the two calls to time.Now().
>>>>>>>>
>>>>>>>> If you substitute this:
>>>>>>>> fmt.Printf("input: %20.18f\n", b)
>>>>>>>>
>>>>>>>> you get something like
>>>>>>>> input: 2.999999965553350911
>>>>>>>>
>>>>>>>> HTH
>>>>>>>> Peter
>>>>>>>>
>>>>>>>> On Fri, May 4, 2018 at 7:26 PM, Andrei Avram <[email protected]
>>>>>>>> > wrote:
>>>>>>>>
>>>>>>>>> Hello everyone,
>>>>>>>>>
>>>>>>>>> Today I ran into a situation that is strange to me. I ran the
>>>>>>>>> following code on two Linux machines (go run floor.go), on two
>>>>>>>>> Windows
>>>>>>>>> ones, and on Go Playground.
>>>>>>>>>
>>>>>>>>> package main
>>>>>>>>>
>>>>>>>>> import (
>>>>>>>>> "time"
>>>>>>>>> "math"
>>>>>>>>> )
>>>>>>>>>
>>>>>>>>> func main() {
>>>>>>>>> datetime := time.Now().Add(time.Hour * 24 * 7 * 4 * 12 * 3)
>>>>>>>>> seconds := -1 * int(time.Now().Sub(datetime).Seconds())
>>>>>>>>> a := 29030400
>>>>>>>>> b := float64(seconds) / float64(a)
>>>>>>>>>
>>>>>>>>> println("input:", b)
>>>>>>>>> println("floor:", math.Floor(b))
>>>>>>>>> }
>>>>>>>>>
>>>>>>>>> On Linux the output is:
>>>>>>>>>
>>>>>>>>> input: +3.000000e+000
>>>>>>>>> floor: *+2.000000e+000*
>>>>>>>>>
>>>>>>>>> On Windows and Playground:
>>>>>>>>>
>>>>>>>>> input: +3.000000e+000
>>>>>>>>> floor: *+3.000000e+000*
>>>>>>>>>
>>>>>>>>> As you can see, on Linux the floor value of float value 3 is
>>>>>>>>> rounded down to 2, while on Windows/Playground it's 3.
>>>>>>>>>
>>>>>>>>> The code was ran with Go 1.10 and 1.10.2.
>>>>>>>>>
>>>>>>>>> The system details of one of the Linux machines, as reported by
>>>>>>>>> "go bug":
>>>>>>>>>
>>>>>>>>> go version go1.10.2 linux/amd64
>>>>>>>>> GOARCH="amd64"
>>>>>>>>> GOBIN=""
>>>>>>>>> GOCACHE="/home/msd/.cache/go-build"
>>>>>>>>> GOEXE=""
>>>>>>>>> GOHOSTARCH="amd64"
>>>>>>>>> GOHOSTOS="linux"
>>>>>>>>> GOOS="linux"
>>>>>>>>> GOPATH="/home/msd/go/"
>>>>>>>>> GORACE=""
>>>>>>>>> GOROOT="/usr/local/go"
>>>>>>>>> GOTMPDIR=""
>>>>>>>>> GOTOOLDIR="/usr/local/go/pkg/tool/linux_amd64"
>>>>>>>>> GCCGO="gccgo"
>>>>>>>>> CC="gcc"
>>>>>>>>> CXX="g++"
>>>>>>>>> CGO_ENABLED="1"
>>>>>>>>> CGO_CFLAGS="-g -O2"
>>>>>>>>> CGO_CPPFLAGS=""
>>>>>>>>> CGO_CXXFLAGS="-g -O2"
>>>>>>>>> CGO_FFLAGS="-g -O2"
>>>>>>>>> CGO_LDFLAGS="-g -O2"
>>>>>>>>> PKG_CONFIG="pkg-config"
>>>>>>>>> GOGCCFLAGS="-fPIC -m64 -pthread -fmessage-length=0
>>>>>>>>> -fdebug-prefix-map=/tmp/go-build304261270=/tmp/go-build
>>>>>>>>> -gno-record-gcc-switches"
>>>>>>>>> GOROOT/bin/go version: go version go1.10.2 linux/amd64
>>>>>>>>> GOROOT/bin/go tool compile -V: compile version go1.10.2
>>>>>>>>> uname -sr: Linux 4.13.0-37-generic
>>>>>>>>> Distributor ID: Ubuntu
>>>>>>>>> Description: Ubuntu 16.04.4 LTS
>>>>>>>>> Release: 16.04
>>>>>>>>> Codename: xenial
>>>>>>>>> /lib/x86_64-linux-gnu/libc.so.6: GNU C Library (Ubuntu GLIBC
>>>>>>>>> 2.23-0ubuntu10) stable release version 2.23, by Roland McGrath et al.
>>>>>>>>> gdb --version: GNU gdb (Ubuntu 7.11.1-0ubuntu1~16.5) 7.11.1
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Is there something I miss or could this be an issue?
>>>>>>>>>
>>>>>>>>> Thanks,
>>>>>>>>> Andrei
>>>>>>>>>
>>>>>>>>> --
>>>>>>>>> You received this message because you are subscribed to the Google
>>>>>>>>> Groups "golang-nuts" group.
>>>>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>>>>> send an email to [email protected].
>>>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>> You received this message because you are subscribed to the Google
>>>>>>> Groups "golang-nuts" group.
>>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>>> send an email to [email protected].
>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>
>>>>>>
>>>>>> --
>>>>> You received this message because you are subscribed to the Google
>>>>> Groups "golang-nuts" group.
>>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>>> an email to [email protected].
>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>
>>>>
>>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "golang-nuts" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to [email protected].
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
--
You received this message because you are subscribed to the Google Groups
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
