BatchtoolsParam handles reduce.in.order differently from others

[DarwinAwardWinner]

Most BPPARAM objects handle the reduce.in.order argument of bpiterate without requiring REDUCE and INIT arguments. However, BatchtoolsParam objects cannot handle this case, which means that some uses of bpiterate can work with some valid BPPARAM arguments but fail with others. This is highly unintuitive. Reprex below:

library(BiocParallel)
library(assertthat)
library(batchtools)

expected_result <- lapply(1:10, sqrt)

assert_that(identical(expected_result, bpiterate(bpiterateAlong(1:10), sqrt, reduce.in.order=TRUE, BPPARAM = SerialParam())))
#> [1] TRUE
assert_that(identical(expected_result, bpiterate(bpiterateAlong(1:10), sqrt, reduce.in.order=TRUE, BPPARAM = MulticoreParam(2))))
#> [1] TRUE
assert_that(identical(expected_result, bpiterate(bpiterateAlong(1:10), sqrt, reduce.in.order=TRUE, BPPARAM = SnowParam(2))))
#> [1] TRUE
## This one fails
assert_that(identical(expected_result, bpiterate(bpiterateAlong(1:10), sqrt, reduce.in.order=TRUE, BPPARAM = BatchtoolsParam(2, cluster = "multicore"))))
#> Error in `.local()`:
#> ! REDUCE must be provided when 'reduce.in.order = TRUE'
## Needs additional arguments to work, although these specific
## arguments may not work in every case
assert_that(identical(expected_result, bpiterate(bpiterateAlong(1:10), sqrt, reduce.in.order=TRUE, BPPARAM = BatchtoolsParam(2, cluster = "multicore"), init = list(), REDUCE = c))) #
#> Submitting 10 jobs in 2 chunks using cluster functions 'Multicore' ...
#> [1] TRUE

^{Created on 2026-05-28 with reprex v2.1.1}

BatchtoolsParam seems to be the odd one out here. Is it possible to bring BatchtoolsParam in line with the other param classes here?

Session info:

> sessionInfo()
R version 4.5.3 (2026-03-11)
Platform: aarch64-apple-darwin24.6.0
Running under: macOS Sequoia 15.7.4

Matrix products: default
BLAS:   /opt/homebrew/Cellar/openblas/0.3.32/lib/libopenblasp-r0.3.32.dylib 
LAPACK: /opt/homebrew/Cellar/r/4.5.3/lib/R/lib/libRlapack.dylib;  LAPACK version 3.12.1

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: America/New_York
tzcode source: internal

attached base packages:
[1] graphics  grDevices utils     datasets  stats     methods   base     

other attached packages:
 [1] batchtools_0.9.18   assertthat_0.2.1    BiocParallel_1.44.0 tidyr_1.3.2         future_1.68.0       devtools_2.4.6      usethis_3.2.1       openxlsx_4.2.8.1    magrittr_2.0.5      dplyr_1.2.1         rex_1.2.1           glue_1.8.1          stringr_1.6.0       ggplot2_4.0.1       colorout_1.3-2     

loaded via a namespace (and not attached):
 [1] gtable_0.3.6       xfun_0.57          remotes_2.5.0      processx_3.8.6     callr_3.7.6        vctrs_0.7.3        tools_4.5.3        pak_0.9.1          ps_1.9.1           generics_0.1.4     base64url_1.4      parallel_4.5.3     tibble_3.3.1       pkgconfig_2.0.3    data.table_1.18.4  checkmate_2.3.4   
[17] RColorBrewer_1.1-3 S7_0.2.2           lifecycle_1.0.5    compiler_4.5.3     farver_2.1.2       progress_1.2.3     codetools_0.2-20   htmltools_0.5.9    snow_0.4-4         yaml_2.3.12        pillar_1.11.1      crayon_1.5.3       ellipsis_0.3.2     cachem_1.1.0       sessioninfo_1.2.3  parallelly_1.45.1 
[33] brew_1.0-10        tidyselect_1.2.1   zip_2.3.3          digest_0.6.39      stringi_1.8.7      purrr_1.2.2        listenv_0.10.0     fastmap_1.2.0      grid_4.5.3         cli_3.6.6          dichromat_2.0-0.1  pkgbuild_1.4.8     clipr_0.8.0        withr_3.0.2        prettyunits_1.2.0  scales_1.4.0      
[49] backports_1.5.1    rappdirs_0.3.4     rmarkdown_2.31     globals_0.18.0     otel_0.2.0         hms_1.1.4          evaluate_1.0.5     memoise_2.0.1      knitr_1.51         rlang_1.2.0        Rcpp_1.1.1-1.1     reprex_2.1.1       pkgload_1.4.1      rstudioapi_0.17.1  R6_2.6.1           fs_2.1.0          

As you can see, I'm not running this test on the latest R/Bioconductor versions, but looking in the code I see that the error message is still there: https://github.com/Bioconductor/BiocParallel/blob/devel/R/BatchtoolsParam-class.R#L465

[DarwinAwardWinner]

I think we can get behavior equivalent to the other param objects by setting init = list() and REDUCE = function(current, new) c(current, list(new)), but I'm not certain. If this is the case, then perhaps we could check if missing(init) && missing(reduce) and then set these defaults?