loadd(deps = TRUE) should work even if the target is missing.

[wlandau]

Prework

• Read and abide by drake's code of conduct.
• Search for duplicates among the existing issues, both open and closed.
• Advanced users: verify that the bug still persists in the current development version (i.e. remotes::install_github("ropensci/drake")) and mention the SHA-1 hash of the Git commit you install.

[wlandau]

This needs to work for debugging. Reprex:

library(drake)
plan <- drake_plan(a = "x", b = a + 1)
make(plan)
#> target a
#> target b
#> fail b
#> Error: Target `b` failed. Call `diagnose(b)` for details. Error message:
#>   non-numeric argument to binary operator
config <- drake_config(plan)
loadd(b, config = config, deps = TRUE)
#> Error in loadd(b, config = config, deps = TRUE): object 'b' not found

^{Created on 2019-04-09 by the reprex package (v0.2.1)}

[wlandau]

Important note: we will need to disable tidyselect functionality when deps is TRUE. For example, loadd(starts_with("model_"), config = config, deps = TRUE)
will not work. For the selection mechanism to work, the model_* targets to need to already be in the cache, which is not always the case when you are debugging your projects. To help drake understand what you mean, you will need to name the targets explicitly when deps is TRUE, e.g. loadd(model_A, model_B, config = config, deps = TRUE).

Also, @kendonB, I am still thinking about #823. I do agree that supplying config is an inconvenience, but none of the alternatives I can think of would scale well. We would at least need to store config$layout$EVERY_TARGET$deps_build$memory or config$graph, which could eat up a lot of space.

[wlandau]

Fixed in a83e7bb. Also added the above note in the loadd() help file under deps, plus an informative console message when the deps and tidyselect arguments are both TRUE.