[Bioc-devel] Packages with broken unit tests because of sample() change in R 3.6

Pages, Herve hp@ge@ @end|ng |rom |redhutch@org
Wed Apr 17 08:53:49 CEST 2019


Hi Leonardo,

Thanks for fixing the derfinder and derfinderHelper tests.

Note that Luke Tierney already suggested the RNGversion("3.5.0") trick 
on this list last month.

IMO using RNGversion("3.5.0") to repair the unit tests broken by this 
change in R 3.6 should be considered a temporary workaround. I would 
still encourage people to update their unit tests to make them work with 
the new RNG. This sounds like a cleaner solution in the long run.

Best,

H.


On 4/10/19 11:29, Leonardo Collado Torres wrote:
> Thanks Hervé,
>
> I also found this tweet by John Muschelli useful for this situation
> https://urldefense.proofpoint.com/v2/url?u=https-3A__twitter.com_StrictlyStat_status_1103303028751372289&d=DwIFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=8-j7ZmUPgzjbPjswR-NgkqVYXyuW5sLTrpbH8t-C-wk&s=YKIxEhvzcorKKg3ZyrD270RZZyQLZZGMKkrQomC8LhA&e= where he
> suggests using suppressWarnings(RNGversion("3.5.0"))
>
> That's what I did for example in
> https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_leekgroup_derfinderHelper_commit_c2c756d6e1f6c3da219073a6c36c2b475453ecf1&d=DwIFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=8-j7ZmUPgzjbPjswR-NgkqVYXyuW5sLTrpbH8t-C-wk&s=cpkWykYsvaXUWf5hpSIUqpoRis480-xt46plnJiHVkk&e=
> to resolve the issue with derfinderHelper.
>
> Best,
> Leo
>
> On Mon, Apr 8, 2019 at 6:40 PM Pages, Herve <hpages using fredhutch.org> wrote:
>> Dear developers,
>>
>> As reported on this list in March (https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.ethz.ch_pipermail_bioc-2Ddevel_2019-2DMarch_014776.html&d=DwIFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=8-j7ZmUPgzjbPjswR-NgkqVYXyuW5sLTrpbH8t-C-wk&s=dQsh_nbcI6Nzrju8GJ0gNCotZieSJ3CXl4Fh6-91-YM&e=) the sample() function has changed in recent R-devel. This change breaks the 19 following Bioconductor software packages:
>>
>>    - cicero
>>    - clipper
>>    - clusterExperiment
>>    - CMA
>>    - CNVPanelizer
>>    - derfinder
>>    - derfinderHelper
>>    - flowQB
>>    - mAPKLA
>>    - MCbiclust
>>    - Melissa
>>    - metaseqR
>>    - nucleoSim
>>    - onlineFDR
>>    - PathNet
>>    - phyloseq
>>    - SPONGE
>>    - treeio
>>    - trena
>>
>>
>> For all these packages (except for the CMA package), the change to sample() breaks the unit tests where sample() is used to generate random data as input to the test. Since sample() generates different random numbers in R 3.6 (compared to R 3.5, with the seed set to the same value), the input to the test now is different and so is its output, so comparison to the old expected result now fails. The fix is to simply modify the expected result to match the new output.
>>
>> If you maintain one of the above package, please fix your package as soon as possible.
>>
>> Keep in mind that all Bioconductor packages will need to pass 'R CMD build' and 'R CMD check' with no error in order to be included in the next Bioconductor release, scheduled for Tuesday April 30. Note that Bioconductor 3.9 will be based on R 3.6 (scheduled to be released on Friday April 26) so please make sure to use a recent version of R devel when testing/fixing your package.
>>
>> See our release schedule:
>>
>>    https://urldefense.proofpoint.com/v2/url?u=https-3A__bioconductor.org_developers_release-2Dschedule_&d=DwIFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=8-j7ZmUPgzjbPjswR-NgkqVYXyuW5sLTrpbH8t-C-wk&s=2ZKn8WEyeS-JlToblg4K59Sj2ffLsGGYAC8B-mD1CA8&e=
>>
>> and this recent announcement on this list:
>>
>>    https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.ethz.ch_pipermail_bioc-2Ddevel_2019-2DApril_014832.html&d=DwIFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=8-j7ZmUPgzjbPjswR-NgkqVYXyuW5sLTrpbH8t-C-wk&s=OF2Yv222QzUT86QMclbwGL7d4M6lbrMDwTqAVxNubY0&e=
>>
>> Thank you for your ongoing contribution to Bioconductor.
>>
>> Cheers,
>>
>> H.
>>
>>
>> --
>> Hervé Pagès
>>
>> Program in Computational Biology
>> Division of Public Health Sciences
>> Fred Hutchinson Cancer Research Center
>> 1100 Fairview Ave. N, M1-B514
>> P.O. Box 19024
>> Seattle, WA 98109-1024
>>
>> E-mail: hpages using fredhutch.org<mailto:hpages using fredhutch.org>
>> Phone:  (206) 667-5791
>> Fax:    (206) 667-1319
>>
>>
>>          [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> Bioc-devel using r-project.org mailing list
>> https://urldefense.proofpoint.com/v2/url?u=https-3A__stat.ethz.ch_mailman_listinfo_bioc-2Ddevel&d=DwIFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=8-j7ZmUPgzjbPjswR-NgkqVYXyuW5sLTrpbH8t-C-wk&s=_pwfoPYdcaP7JeYK6uuQ7ppova4vVpoyM6Bunls422U&e=

-- 
Hervé Pagès

Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M1-B514
P.O. Box 19024
Seattle, WA 98109-1024

E-mail: hpages using fredhutch.org
Phone:  (206) 667-5791
Fax:    (206) 667-1319



More information about the Bioc-devel mailing list