Skip to content

[R-pkg-devel] Unreproducible error for CRAN submission on winbuilder using r-devel

4 messages · Uwe Ligges, Duncan Murdoch, Chris Brien

Chris Brien
# 
Hi all,

I submitted my package asremlPlus 4.4.47 to CRAN on Thurs 3/04/2025 and it generated an Error (and a NOTE) on Winbuilder using R version 4.5.0 RC (2025-04-04 r88102 ucrt):

package asremlPlus_4.4.47.tar.gz does not pass the incoming checks automatically, please see the following pre-tests (additional issue checks):
Windows: <https://win-builder.r-project.org/incoming_pretest/asremlPlus_4.4.47_20250403_120518/Windows/00check.log<https://win-builder.r-project.org/incoming_pretest/asremlPlus_4.4.47_20250403_120518/Windows/00check.log>>
Status: 1 ERROR, 1 NOTE

I ran R CMD check on the package using r-devel r88102 and was unable to generate the ERROR.

So, I submitted the package to Winbuilder r-devel and the package does not generate an error using R version 4.5.0 RC (2025-04-04 r88112 ucrt) - see https://win-builder.r-project.org/Ht2zyDE9N1Xx/00check.log.

Can I assume that the error was transitory and resubmit my package to CRAN?

Cheers,

 Chris
Uwe Ligges
# 
This is on so far strange as the winbuilder service and the CRAN 
incoming check happen actually on the same machine, same library and 
same version of R.

If this is not related to some randomness in generating your data, 
please simply resubmit.

Best,
Uwe Ligges
On 06.04.2025 10:08, Chris Brien wrote:
Duncan Murdoch
#
On 2025-04-06 4:08 a.m., Chris Brien wrote:
The error was

"Error in pf(q = Fvalue, df1 = ndf, df2 = ddf, lower.tail = FALSE) :
   Non-numeric argument to mathematical function"

That doesn't look like something that would be transitory.  If you are 
working with random data it might be hard to reproduce, but still worth 
worrying about.  Or it could be transitory because some package you're 
using was updated between the two runs, or some weird timing issue 
caused some value to be wrong.

Duncan Murdoch
Chris Brien
# 
Hi Duncan,

Thanks for your comments.

"Error in pf(q = Fvalue, df1 = ndf, df2 = ddf, lower.tail = FALSE) :
Non-numeric argument to mathematical function"
Indeed, and the problem arises from a call to the anova function of lmerTest whose latest versions is 2020-10-23.


That doesn't look like something that would be transitory. If you are
working with random data it might be hard to reproduce, but still worth
worrying about. Or it could be transitory because some package you're
using was updated between the two runs, or some weird timing issue
caused some value to be wrong.

The data is a data set from asremlplus called Oat.dat, so not random at all.

It seems that the most likely explanation is that it is some weird transitory issue on the CRAN incoming check that caused some value to be wrong.

Anyway, asremlPlus has now been resubmitted and is on its way to CRAN.

Cheers,

 Chris



From: Duncan Murdoch <murdoch.duncan at gmail.com>
Sent: Sunday, 6 April 2025 11:03 PM
To: Chris Brien <chris.brien at adelaide.edu.au>; r-package-devel at r-project.org
Subject: Re: [R-pkg-devel] Unreproducible error for CRAN submission on winbuilder using r-devel
On 2025-04-06 4:08 a.m., Chris Brien wrote:
The error was

"Error in pf(q = Fvalue, df1 = ndf, df2 = ddf, lower.tail = FALSE) :
Non-numeric argument to mathematical function"

That doesn't look like something that would be transitory. If you are
working with random data it might be hard to reproduce, but still worth
worrying about. Or it could be transitory because some package you're
using was updated between the two runs, or some weird timing issue
caused some value to be wrong.

Duncan Murdoch