[R-pkg-devel] crowdsourcing/help with

[Ben Bolker]

   The win-builder output is here: 
https://win-builder.r-project.org/ce55MsIkho3o/00check.log

   Key points:

  the Anova.glmmTMB example fails, but this is the only output provided:

 >
 > ### ** Examples
 >
 > warp.lm <- glmmTMB(breaks ~ wool * tension, data = warpbreaks)
 > salamander1 <- 
up2date(readRDS(system.file("example_files","salamander1.rds",package="glmmTMB")))

  (I was wrong that examples fail immediately: 
https://win-builder.r-project.org/ce55MsIkho3o/examples_and_tests/glmmTMB-Ex.Rout 
shows that some examples do succeed)

    I will try to make a test-A.R that runs first that *only* runs 
up2date (which is pretty much just a call to TMB::MakeADFun).  I might 
try a call to library(TMB), which might (?) throw an informative warning 
about a Matrix/TMB binary mismatch ...

   thanks
   Ben



On 3/10/22 3:19 AM, Tomas Kalibera wrote:
> 
> On 3/10/22 01:38, Ben Bolker wrote:
>>
>>   We (the glmmTMB developers) are having a hard-to-reproduce problem 
>> on Windows.  When run on CRAN's check platform or win-builder with 
>> r-devel, the code fails on the first example, when trying to 
>> reconstruct a 'TMB' object (TMB is a C++-based package we depend on). 
>> The problem is that we can't find a Windows system *other* than 
>> win-builder on which it fails -- it works on r-hub, and with 2 or 3 
>> other Windows 10/11 systems we've tried it on.
>>
>>   I suspect this has to do with an incompatibility in the binary stack 
>> - there is a dependency chain Matrix → TMB → glmmTMB, and we know that 
>> weird stuff can happen if the stack is incompatible (e.g. TMB was 
>> built with an older version of Matrix than is currently installed), 
>> but haven't managed to put together a stack that fails.
>>
>>   If anyone with a Windows system handy would be willing to try this 
>> out and see if it works that would be *super* handy (especially if you 
>> can get it to fail!)
>>
>>   You can read all about it here:
>>
>> https://github.com/glmmTMB/glmmTMB/issues/787
>>
>>    If you just want to grab the tarball so you can run R CMD check on 
>> it, try:
>>
>> r <- remotes::remote_download(github_remote("glmmTMB/glmmTMB/glmmTMB"))
>> file.rename(r, "my_glmmtmb_tarball.tar.gz")
>>
>>    If you do manage to get it to fail, it would be great to see 
>> detailed output from
>>
>> library(glmmTMB)
>> up2date(readRDS(system.file("example_files","salamander1.rds",package="glmmTMB"))) 
>>
>>
>>   Any other debugging/diagnostic suggestions are of course welcome.
>>   (If this fails we will probably have to put "skip if on Windows" 
>> logic around most of our examples and tests.)
> 
> As you have a machine where it fails, which is winbuilder, could you 
> perhaps instrument your package to print the information you need to 
> debug the issue, and re-submit it? Also, to get more specific advice, 
> perhaps you could paste the actual output or a link to it.
> 
> Best
> Tomas
> 
>>
>>   cheers
>>     Ben Bolker
>>
>> ______________________________________________
>> R-package-devel using r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-package-devel

-- 
Dr. Benjamin Bolker
Professor, Mathematics & Statistics and Biology, McMaster University
Director, School of Computational Science and Engineering
(Acting) Graduate chair, Mathematics & Statistics