[R-pkg-devel] some notes

[Spencer Graves]

On 2018-04-24 09:50, Duncan Murdoch wrote:
> On 24/04/2018 10:29 AM, Xubo Yue wrote:
>> Hey
>>
>> I just got feedback from CRAN auto-checking and here are some notes I
>> cannot resolve:
>>
>> * checking CRAN incoming feasibility ... NOTE
>> Maintainer: 'Jian Kang <jiankang at umich.edu>'
>>
>> New submission
>>
>> Possibly mis-spelled words in DESCRIPTION:
>>    Kang (7:196)
>>
>> I am not sure how to solve this.
>>
>
> You can ignore those.
>
>> Also, here are some running time notes:
>>
>> ** running examples for arch 'i386' ... [250s] NOTE
>> Examples with CPU or elapsed time > 10s
>>                            user system elapsed
>> multipleRegion_plot     190.35   1.92  205.61
>> summary_kcca              1.40   0.17   14.79
>> permkCCA_multipleRegion   1.19   0.11   14.51
>> nii2RData                 0.40   0.20   13.84
>> ** running examples for arch 'x64' ... [79s] NOTE
>> Examples with CPU or elapsed time > 10s
>>                           user system elapsed
>> multipleRegion_plot     18.36   2.06   33.66
>> permkCCA_multipleRegion  1.47   0.20   14.88
>> summary_kcca             1.48   0.16   14.85
>> nii2RData                0.62   0.23   14.09
>> * checking PDF version of manual ... OK
>> * DONE
>> Status: 3 NOTEs
>>
>> My total running time is less than 10 minutes. I am wondering is it 
>> ok to
>> ignore these two NOTEs?
>
> You shouldn't ignore those.  Your package will be tested many, many 
> times, and it will take more than 3 minutes for just one example.  (It 
> will be tested whenever any package you depend on is changed, to make 
> sure the change doesn't break yours.)
>
> Use a smaller dataset, do less, or whatever you can to get the 
> multipleRegion_plot example down to a time similar to the other ones, 
> i.e. well below 10s in user+system time.


       If it becomes infeasible to find smaller datasets, etc., you can 
cut out certain tests with a construct like the following:


if(!fda::CRAN()){
# ... tests to run on other computers but to skip on CRAN.  ...
}


       Obviously, you'll need fda::CRAN.  I found it infeasible to meet 
that requirement and so disabled the tests that violated CRAN's rule on 
this.


       I've been told that this is a suboptimal solution and should be 
avoided.  I agree.


       Hope this helps.
       Spencer Graves

>
> You don't have a lot of control of elapsed time because it will depend 
> on what else is happening on the test machine, but if the other times 
> are short, usually elapsed time will also be short.
>
> Duncan Murdoch
>
> ______________________________________________
> R-package-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-package-devel