[R-pkg-devel] unusually slow tests on win-builder

Jonathon Love jon at thon.cc
Tue Jun 6 04:04:04 CEST 2017


> So your windows installation of R-3.4.0 can solve this in 600ms? Or is 
> yoiur another OS?

ah, yes, another OS. it turns out on windows this takes 6 seconds. the 
issue is that we only load imports on demand (not in our NAMESPACE), 
there's a behemoth dependency chain that that analysis invokes, and 
package loading appears to be *quite a bit* slower on windows than 
macOS/linux.

with thanks

jonathon

> For winbuilder, it could be it was heavily loaded, but even then it 
> should never take more than twice as long as on an idling machine. 
> Then winbuilder has server CPUs with lots of cores hence operates at a 
> lower frequency, altogether not more than 3-4 times the the time you 
> got can be expected.
>
> Best,
> Uwe Ligges
>
>
>
>>
>> any other suggestions as to what might be going on here?
>>
>> with thanks
>>
>> jonathon
>>
>>
>> On 5/6/17 22:36, Dason Kurkiewicz wrote:
>>> The relevant output is below:
>>>
>>> * checking examples ...
>>> ** running examples for arch 'i386' ... [24s] NOTE
>>> Examples with CPU or elapsed time > 10s
>>>      user system elapsed
>>> cfa 8.05   0.36   10.08
>>> ** running examples for arch 'x64' ... [28s] NOTE
>>> Examples with CPU or elapsed time > 10s
>>>       user system elapsed
>>> cfa 10.62   0.23      12
>>> * checking for unstated dependencies in 'tests' ... OK
>>> * checking tests ...
>>> ** running tests for arch 'i386' ... [25s] OK
>>>    Running 'testthat.R' [24s]
>>> ** running tests for arch 'x64' ... [32s] OK
>>>    Running 'testthat.R' [31s]
>>>
>>>
>>> I'm not sure why you're saying it's taking too long to run your tests
>>> when the output is telling you that it's your examples that are taking
>>> too long to run.  It's also not taking 24 and 28 seconds - I believe
>>> that is just a timestamp telling you that it's producing that note 24
>>> and 28 seconds into the check but I could be wrong.  Either way the
>>> relevant output for how long the examples take look like:
>>>
>>>      user system elapsed
>>> cfa 8.05   0.36   10.08
>>>
>>> That which tells you total it took 10.08 seconds to run the examples
>>> for arch 'i386'.
>>>
>>> -Dason Kurkiewicz
>>>
>>> On Mon, Jun 5, 2017 at 8:23 AM, Jonathon Love <jon at thon.cc> wrote:
>>>> hi,
>>>>
>>>> i've uploaded a version of my package to the win-builder service, 
>>>> but it is
>>>> reporting that one of my tests takes > 10 seconds to run. examining 
>>>> the log
>>>> here:
>>>>
>>>> https://win-builder.r-project.org/BvR4nreDE9qP/00check.log
>>>>
>>>> the win-builder is taking 24 and 28 seconds to run this test.
>>>>
>>>> on my local machine, this same test runs in 600ms
>>>>
>>>> can anyone shed any light on this? why is the win-builder so slow?
>>>>
>>>> with thanks
>>>>
>>>> jonathon
>>>>
>>>> ______________________________________________
>>>> R-package-devel at r-project.org mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/r-package-devel
>>
>> ______________________________________________
>> R-package-devel at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-package-devel



More information about the R-package-devel mailing list