Improve type stability of dynamic branching

[wlandau]

Summary

This PR improves the type stability of dynamic branching. It changes the return values from group() dynamic targets and from readd(), so this is technically a breaking change. However, dynamic branching is still new and this change is important enough that a little discomfort is worth it in the end. Type stability is too important, as I discuss in #1087 (comment) and #1105 (comment).

Previously, readd() always returned a list for dynamic targets, as I demonstrated in #1105 (comment). Now, it returns the result of vec_c() which respects the types of the sub-targets.

library(drake)
plan <- drake_plan(
  x = seq_len(2),
  y = mtcars[x, ],
  z = target(y, dynamic = group(y, .by = x))
)
make(plan)
#> target x
#> target y
#> dynamic z
#> subtarget z_25909f33
#> subtarget z_7fb9021c
#> aggregate z

readd(z)
#>   mpg cyl disp  hp drat    wt  qsec vs am gear carb
#> 1  21   6  160 110  3.9 2.620 16.46  0  1    4    4
#> 2  21   6  160 110  3.9 2.875 17.02  0  1    4    4

^{Created on 2019-12-10 by the reprex package (v0.3.0)}

Similarly, group()'ed dynamic dependencies were always lists before. Now, they respect the types of the sub-targets.

library(drake)
plan <- drake_plan(
  x = c(1, 1, 2, 2),
  y = mtcars[seq_len(4), ],
  z = target(y, dynamic = group(y, .by = x))
)
make(plan)
#> In drake, consider r_make() instead of make(). r_make() runs make() in a fresh R session for enhanced robustness and reproducibility.
#> target x
#> target y
#> dynamic z
#> subtarget z_6f19200e
#> subtarget z_e9abea21
#> aggregate z

readd(subtargets(z)[1], character_only = TRUE)
#>               mpg cyl disp  hp drat    wt  qsec vs am gear carb
#> Mazda RX4      21   6  160 110  3.9 2.620 16.46  0  1    4    4
#> Mazda RX4 Wag  21   6  160 110  3.9 2.875 17.02  0  1    4    4

readd(z)
#>    mpg cyl disp  hp drat    wt  qsec vs am gear carb
#> 1 21.0   6  160 110 3.90 2.620 16.46  0  1    4    4
#> 2 21.0   6  160 110 3.90 2.875 17.02  0  1    4    4
#> 3 22.8   4  108  93 3.85 2.320 18.61  1  1    4    1
#> 4 21.4   6  258 110 3.08 3.215 19.44  1  0    3    1

^{Created on 2019-12-10 by the reprex package (v0.3.0)}

In some ways, we need to be more careful about how we aggregate targets. drake will no longer automatically compartmentalize sub-targets in lists for us. When we aggregate sub-targets, we may get a value we do not expect (i.e. an unexpectedly long vector) but I think this PR is still the right way to go.

library(drake)
plan <- drake_plan(
  w = c(1, 1, 2, 2),
  x = letters[seq_len(4)],
  y = LETTERS[seq_len(4)],
  z = target(c(x, y), dynamic = group(x, y, .by = w))
)
make(plan)
#> target w
#> target x
#> target y
#> dynamic z
#> subtarget z_bf3d8724
#> subtarget z_b9172820
#> aggregate z

readd(z)
#> [1] "a" "b" "A" "B" "c" "d" "C" "D"

readd(subtargets(z)[1], character_only = TRUE)
#> [1] "a" "b" "A" "B"

^{Created on 2019-12-10 by the reprex package (v0.3.0)}

Related GitHub issues and pull requests

• Ref: Fully embrace vctrs for dynamic branching #1105

Checklist

• I understand and agree to drake's code of conduct.
• I have listed any substantial changes in the development news.
• I have added testthat unit tests to tests/testthat for any new functionality.
• This pull request is not a draft.

[codecov-io]

Codecov Report

Merging #1106 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff           @@
##           master   #1106   +/-   ##
======================================
  Coverage     100%    100%           
======================================
  Files          47      47           
  Lines        7424    7413   -11     
======================================
- Hits         7424    7413   -11

Impacted Files	Coverage Δ
R/drake_plan.R	100% <ø> (ø)	⬆️
R/drake_plan_helpers.R	100% <ø> (ø)	⬆️
R/make.R	100% <ø> (ø)	⬆️
R/dynamic.R	100% <100%> (ø)	⬆️
R/manage_memory.R	100% <100%> (ø)	⬆️
R/cache.R	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1518467...1966ac3. Read the comment docs.