CLEANUP: Deprecation of multiprocess (since Oct 2020)

[HenrikBengtsson]

Packages on CRAN and Bioconductor that use 'multiprocess'

Last updated: 2023-03-21

Currently, multiprocess is defunct, but still exists. When multiprocess is removed in a later version, it will break 3 CRAN packages (and zero Bioconductor packages):

• [ ] PeakSegDisk

  • [ ] Code: https://github.com/tdhock/PeakSegDisk/search?q=multiprocess&type=code and https://github.com/cran/PeakSegDisk/search?q=multiprocess
  • [x] Issue: https://github.com/tdhock/PeakSegDisk/issues/13 (2022-08-28, 2023-01-17, 2023-03-07)
  • [ ] On CRAN
  • Number of direct reverse dependencies: 1 (soft)
  • Emailed affected revdep maintainers: N/A (same maintainer)
  • Notes:
    • Will only give an error in interactive mode, because it uses if(interactive() && requireNamespace("future"))future::plan("multiprocess").
    • Because it uses "multiprocess", it will not break when the multiprocess function is removed.
    • Conclusions: This is a non-issue for me when updating future.

• [ ] tcplfit2

  • [ ] Code: https://github.com/cran/tcplfit2/search?q=multiprocess
  • [x] BugReport: by email on 2021-10-18. Reminder on 2022-09-15,, 2023-01-17
  • [ ] On CRAN
  • Number of direct reverse dependencies: 1 (hard)
  • Emailed affected revdep maintainers: N/A (same maintainer)
  • Notes:
  • Will only fail if tcplfit2::hitcont() is called with mc.cores > 1. Uses plan(multiprocess, workers = mc.cores).
  • Package uses import(future), so when multiprocess is removed from future, this package will still load, but R CMD check will produce a NOTE (or a WARNING?).

• [ ] TSstudio

  • [ ] Code: https://github.com/RamiKrispin/TSstudio/search?q=multiprocess
  • [ ] https://github.com/RamiKrispin/TSstudio/issues/63 (2021-10-18, 2022-09-27, 2023-01-17) Blocker: maintainer cannot update due to contract with their employer (2022-04-12)
  • [ ] On CRAN
  • Number of direct reverse dependencies: 1 (soft)
  • Emailed affected revdep maintainers: N/A (same maintainer)
  • Notes:
  • Not sure why it's not failing its checks now. Function ts_grid() uses future::plan(future::multiprocess, workers = n.cores) which is called when parallel = TRUE, which is the default.

If you see this and are a maintainer of one of the above packages, please replace multiprocess with multisession. The multiprocess backend is defunct as of future 1.32.0 (2023-03-07), i.e. give an error if used.

Additional non-CRAN, non-Bioconductor packages

• [ ] varimpact
  • [ ] Code: https://github.com/ck37/varimpact/search?q=multiprocess
  • [x] https://github.com/ck37/varimpact/issues/22 (2021-11-22)

Completed

Last updated 2023-03-21.

Packages that fail checks because 'multiprocess' is now defunct

future 1.32.0 is on CRAN as of 2023-03-07. This broke 3 CRAN packages and 1 Bioconductor package.

• [x] dhReg

  • [ ] Code: https://github.com/cran/dhReg/search?q=multiprocess
  • [x] BugReport: by email on 2021-10-18, 2022-09-27, 2023-01-17, 2023-02-12
  • [x] Removed from CRAN on 2023-03-21
  • Number of direct reverse dependencies: 0

• [x] fiery

  • [ ] Source code: https://github.com/search?q=repo%3Athomasp85%2Ffiery%20multiprocess&type=code
  • [ ] Issue: https://github.com/thomasp85/fiery/issues/50 (2023-02-01, 2023-02-14)
  • [x] On CRAN
  • Number of direct reverse dependencies: 2 (soft)
  • Emailed affected revdep maintainers: N/A (same maintainer)

• [x] prewas

  • [ ] Code: https://github.com/cran/prewas/search?q=multiprocess
  • [x] Already done (cf. https://github.com/Snitkin-Lab-Umich/prewas/commit/0bf01d3c63e8ea0e5306adc9c9620a2702b5329f) but not yet on CRAN
  • [x] https://github.com/Snitkin-Lab-Umich/prewas/issues/13 (2021-10-18, 2022-09-15, 2023-01-17, 2023-02-12)
  • [x] Removed from CRAN on 2023-03-21
  • [ ] Number of direct reverse dependencies: 0

• [ ] MineICA

  • [ ] Code: https://code.bioconductor.org/browse/MineICA/blob/master/R/runAn.R?branch=master&file=R/runAn.R#L127
  • [ ] BugReport URL: Maintainer emailed (2023-02-13)
  • [ ] On Bioconductor devel
  • [ ] On Bioconductor release
  • Number of direct reverse dependencies: 0

Resolved

• [x] AlpsNMR

  • [x] Code: https://github.com/sipss/AlpsNMR/search?q=multiprocess
  • [x] https://github.com/sipss/AlpsNMR/issues/65 (2022-05-02)
  • [x] On Bioconductor devel 3.16 (to become release in Oct 2022)
  • [x] On Bioconductor release 3.15

• [x] alookr

  • [x] Code: https://github.com/choonghyunryu/alookr/search?q=multiprocess
  • [x] https://github.com/choonghyunryu/alookr/issues/2
  • [x] On CRAN (2022-06-12)

• [x] blockCV

  • [ ] https://github.com/rvalavi/blockCV/issues/23 (2022-04-27, 2022-09-29, 2023-01-17, 2023-02-01)
  • [x] On CRAN (2023-02-05)
  • Number of direct reverse dependencies: 2 (hard) + 4 (soft)
  • Emailed affected revdep maintainers: 2023-01-26
  • Resolution: Package removed dependency on future.

• [x] flowGraph

  • [x] Code: https://github.com/aya49/flowGraph/search?q=multiprocess
  • [x] https://github.com/aya49/flowGraph/issues/1 (2022-04-07)
  • [x] On Bioconductor release 3.15 (April 2022)

• [x] geocmeans

  • [x] Code: https://github.com/JeremyGelb/geocmeans/search?q=multiprocess
  • [x] https://github.com/JeremyGelb/geocmeans/issues/3
  • [x] On CRAN

• [x] greta

  • [x] Code: https://github.com/greta-dev/greta/search?q=multiprocess
  • [x] https://github.com/greta-dev/greta/issues/394
  • [x] On CRAN (2022-09-08)

• [x] ipc

  • [x] Code: https://github.com/fellstat/ipc/search?q=multiprocess
  • [x] https://github.com/fellstat/ipc/issues/24 (2022-04-28 and 2022-10-03, 2023-01-17)
  • [x] On CRAN (2023-01-26)
  • Number of direct reverse dependencies: 1 (hard)

• [x] kernelboot

  • [x] Code: https://github.com/twolodzko/kernelboot/search?q=multiprocess
  • [x] https://github.com/twolodzko/kernelboot/issues/9
  • [x] https://github.com/twolodzko/kernelboot/issues/9#issuecomment-975951282
  • [x] On CRAN (2022-09-04)

• [x] protti

  • [x] Code: https://github.com/jpquast/protti/search?q=multiprocess
  • [x] https://github.com/jpquast/protti/issues/160
  • [x] On CRAN

• [x] rBiasCorrection

  • [x] Code: https://github.com/kapsner/rBiasCorrection/search?q=multiprocess
  • [x] https://github.com/kapsner/rBiasCorrection/issues/11
  • [x] On CRAN (2022-06-20)

• [x] RTransferEntropy

  • [x] Code: https://github.com/BZPaper/RTransferEntropy/search?q=multiprocess
  • [x] https://github.com/BZPaper/RTransferEntropy/issues/57 (2022-04-28, 2022-10-19, 2023-01-17)
  • [x] On CRAN (2023-02-01)
  • Number of direct reverse dependencies: 1 (hard)
  • Emailed affected revdep maintainers: 2023-01-26

• [x] seer

  • [x] Code: https://github.com/thiyangt/seer/search?q=multiprocess
  • [x] https://github.com/thiyangt/seer/issues
  • [x] On CRAN

• [x] SpaDES.core

  • [x] Code: https://github.com/PredictiveEcology/SpaDES.core/search?q=multiprocess
  • [x] https://github.com/PredictiveEcology/SpaDES.core/issues/181
  • [x] On CRAN

• [x] tsfeatures

  • [x] Code: https://github.com/cran/tsfeatures/search?q=multiprocess
  • [x] https://github.com/robjhyndman/tsfeatures/issues/44/ (2021-10-18, 2022-10-03)
  • [x] On CRAN

Moot, because removed from CRAN or Bioconductor (last updated 2022-06-02)

• [x] biotmle

  • [ ] Code: https://github.com/nhejazi/biotmle/search?q=multiprocess
  • Comment: No longer depending on future.

• [x] EpiNow2

  • [x] Code: https://github.com/epiforecasts/EpiNow2/search?q=multiprocess (2022-02-18)
  • [ ] Plan on releasing EpiNow2 1.3.3 (2022-09-14)

• [x] fipe

  • [ ] Code: https://github.com/italocegatta/fipe/search?q=multiprocess
  • [x] https://github.com/italocegatta/fipe/issues
  • [x] Removed from CRAN (2022-04-24)

[HenrikBengtsson]

The next step in deprecating 'multiprocess' could be to make it become the same as 'sequential' regardless of argument workers. This will hopefully lead to more developers to update their code sooner.

[zeehio]

You can remove AlpsNMR from this list, since I backported the deprecation fix to the release branch (and it's released already)

[HenrikBengtsson]

Starting with next release of future (>= 1.28.0-9001), there will be one deprecation warning produced per 'multiprocess' future created. This may be very noisy, but that's on purpose. For example,

> library(future)
> plan(multiprocess)
Warning message:
Strategy 'multiprocess' is deprecated in future (>= 1.20.0) [2020-10-30]. Instead, explicitly specify either 'multisession' (recommended) or 'multicore'. In the current R session, 'multiprocess' equals 'multicore'. 

> y <- future.apply::future_lapply(1:3, FUN = sqrt)
Warning messages:
1: Detected creation of a 'multiprocess' future. Strategy 'multiprocess' is deprecated in future (>= 1.20.0) [2020-10-30]. Instead, explicitly specify either 'multisession' (recommended) or 'multicore'. In the current R session, 'multiprocess' equals 'multicore'. 
2: Detected creation of a 'multiprocess' future. Strategy 'multiprocess' is deprecated in future (>= 1.20.0) [2020-10-30]. Instead, explicitly specify either 'multisession' (recommended) or 'multicore'. In the current R session, 'multiprocess' equals 'multicore'. 
3: Detected creation of a 'multiprocess' future. Strategy 'multiprocess' is deprecated in future (>= 1.20.0) [2020-10-30]. Instead, explicitly specify either 'multisession' (recommended) or 'multicore'. In the current R session, 'multiprocess' equals 'multicore'. 
> 

[HenrikBengtsson]

UPDATE 2023-01-17: Next step is to turn 'multiprocess' into 'sequential' processing; still with those warnings

[HenrikBengtsson]

I've made 'multiprocess' defunct when running interactive mode, while keeping it deprecated in batch mode (e.g. R CMD check) for a tad longer;

• Deprecated plan(multiprocess, ...) is now defunct when running in interactive mode. The next step is to make it defunct also when running in batch mode.

[HenrikBengtsson]

I've decided to make multiprocess defunct in the next release, which I'll submit to CRAN today or tomorrow.

[HenrikBengtsson]

future 1.32.0 is on CRAN, where 'multiprocess' is now defunct. The next step will be to remove multiprocess from the API. When multiprocess is removed in a later version, it will break 3 CRAN packages (and zero Bioconductor packages):

• [ ] PeakSegDisk

  • [ ] Code: https://github.com/tdhock/PeakSegDisk/search?q=multiprocess&type=code and https://github.com/cran/PeakSegDisk/search?q=multiprocess
  • [x] Issue: https://github.com/tdhock/PeakSegDisk/issues/13 (2022-08-28, 2023-01-17)
  • [ ] On CRAN
  • Number of direct reverse dependencies: 1 (soft)
  • Emailed affected revdep maintainers: N/A (same maintainer)
  • Note: Will only give an error in interactive mode, because it uses if(interactive() && requireNamespace("future"))future::plan("multiprocess").

• [ ] tcplfit2

  • [ ] Code: https://github.com/cran/tcplfit2/search?q=multiprocess
  • [x] BugReport: by email on 2021-10-18. Reminder on 2022-09-15,, 2023-01-17
  • [ ] On CRAN
  • Number of direct reverse dependencies: 1 (hard)
  • Emailed affected revdep maintainers: N/A (same maintainer)
  • Note: Will only fail if tcplfit2::hitcont() is called with mc.cores > 1. Uses plan(multiprocess, workers = mc.cores).

• [ ] TSstudio

  • [ ] Code: https://github.com/RamiKrispin/TSstudio/search?q=multiprocess
  • [ ] https://github.com/RamiKrispin/TSstudio/issues/63 (2021-10-18, 2022-09-27, 2023-01-17) Blocker: maintainer cannot update due to contract with their employer (2022-04-12)
  • [ ] On CRAN
  • Number of direct reverse dependencies: 1 (soft)
  • Emailed affected revdep maintainers: N/A (same maintainer)
  • Note: Not sure why it's not failing its checks now. Function ts_grid() uses future::plan(future::multiprocess, workers = n.cores) which is called when parallel = TRUE, which is the default.

I'm gonna close this issue and deal with the above at a later time.