Fixing sim_fixed_n() parallel results return

[cmansch]

This update addresses two open issues: #250 and #251.

In order to return the results from each %dofuture% loop, the last object must return some value. The setDF() call previously used resulted in the results not being passed back from the loop when run with a parallel backend. This error was hidden in the results as the loop used the .errorhandling = "pass" instead of "stop".

The parallel vignette was also updated to indicate that sim_gs_n() supports the same backend implementation as sim_fixed_n().

[jdblischak]

I get an error when running it locally. Am I the only one?

git2r::commits(n = 1)
## [[1]]
## [443a59c] 2024-05-28: Updating the parallel vignette to indicate that `sim_gs_n()` also supports the `future` backends.
devtools::load_all()
library("future")
plan("multisession", workers = 2)
sim_fixed_n(n_sim = 10)
## Using 2 cores with backend multisession
## Error in doAnalysis(d, rho_gamma, n_stratum) :
##   could not find function "doAnalysis"
sessionInfo()
## R version 4.3.3 (2024-02-29 ucrt)
## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 11 x64 (build 22631)
## 
## Matrix products: default
## 
## 
## locale:
##   [1] LC_COLLATE=English_United States.utf8  LC_CTYPE=English_United States.utf8   
## [3] LC_MONETARY=English_United States.utf8 LC_NUMERIC=C                          
## [5] LC_TIME=English_United States.utf8    
## 
## time zone: America/New_York
## tzcode source: internal
## 
## attached base packages:
##   [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
##   [1] future_1.33.2    simtrial_0.4.1.3 testthat_3.2.1.1
## 
## loaded via a namespace (and not attached):
##   [1] lattice_0.22-6      stringi_1.8.4       listenv_0.9.1       digest_0.6.35      
## [5] magrittr_2.0.3      grid_4.3.3          iterators_1.0.14    mvtnorm_1.2-4      
## [9] pkgload_1.3.4       fastmap_1.1.1       Matrix_1.6-5        foreach_1.5.2      
## [13] rprojroot_2.0.4     pkgbuild_1.4.4      sessioninfo_1.2.2   brio_1.1.5         
## [17] survival_3.6-4      urlchecker_1.0.1    promises_1.3.0      purrr_1.0.2        
## [21] codetools_0.2-20    cli_3.6.2           shiny_1.8.1.1       rlang_1.1.3        
## [25] parallelly_1.37.1   future.apply_1.11.2 splines_4.3.3       ellipsis_0.3.2     
## [29] remotes_2.5.0       withr_3.0.0         cachem_1.0.8        devtools_2.4.5     
## [33] tools_4.3.3         parallel_4.3.3      memoise_2.0.1       doFuture_1.0.1     
## [37] httpuv_1.6.15       globals_0.16.3      vctrs_0.6.5         R6_2.5.1           
## [41] mime_0.12           lifecycle_1.0.4     git2r_0.33.0        stringr_1.5.1      
## [45] fs_1.6.4            htmlwidgets_1.6.4   usethis_2.2.3       miniUI_0.1.1.1     
## [49] desc_1.4.3          later_1.3.2         glue_1.7.0          data.table_1.15.4  
## [53] profvis_0.3.8       Rcpp_1.0.12         rstudioapi_0.16.0   xtable_1.8-4       
## [57] htmltools_0.5.8.1   compiler_4.3.3

[jdblischak]

Hmmm...the problem appears to be with devtools::load_all()

git2r::commits(n = 1)
## [[1]]
## [443a59c] 2024-05-28: Updating the parallel vignette to indicate that `sim_gs_n()` also supports the `future` backends.
library("simtrial")
library("future")
plan("multisession", workers = 2)
sim_fixed_n(n_sim = 10)
# works

[jdblischak]

I think I figured it out. In short, you can't use devtools::load_all() on Windows to test parallel code that uses "multisession". You can only use devtools::load_all() on unix-alike with "multiprocess" or "multicore. See this Issue for all the details futureverse/future#206

[jdblischak]

One minor nit: since this fixes a known bug that is affecting end users, it would be useful to bump the version number in DESCRIPTION to 0.4.1.4 so that users can confirm they have the fix installed

[cmansch]

One minor nit: since this fixes a known bug that is affecting end users, it would be use to bump the version number in DESCRIPTION to 0.4.1.4 so that users can confirm they have the fix installed

Added a commit to address the version bump!

[LittleBeannie]

Thanks for timely address it, @cmansch !