[ideas] Clarifications about the iteration argument

[multimeric]

Help

• I understand and agree to https://books.ropensci.org/targets/help.html.

Description

I just hit an issue very similar to Elio's discussion here: #1216. The current wording for iteration in the book seems to be:

The iteration argument of tar_target() determines how to split non-dynamic targets and how to aggregate dynamic ones.

I guess in Elio's example paths_list is the non-dynamic target that's being split, and hence its own iteration argument affects how it is split.

Some of my questions about this argument that I'd love to have answered are:

• Does iteration affect the splitting/aggregation of the current target, or incoming targets? In Elio's example it seems to affect the current target. Is that always the case?
• Does iteration only come into effect at the boundary between a dynamic and static target? What about if one dynamic branched target is the input for another dynamic branched target?

Once this is clarified, I'm happy to submit a PR to improve the docs here.

[Aariq]

I was also confused initially about which targets iteration affects, but I think I maybe get it now, and my mental model of what the argument does has changed a bit. I now think of it as basically "how should we treat this target—as a list or as a vector (or a grouped dataframe) when it needs to get split or its pieces need to be recombined?"

[multimeric]

Thanks, that certainly helps with my first question!

[wlandau]

Does iteration only come into effect at the boundary between a dynamic and static target? What about if one dynamic branched target is the input for another dynamic branched target?

In that case, the downstream dynamic target always branches over the existing branch targets of the upstream target. Since the upstream branch targets already exist and do not need to be combined, the iteration argument of the upstream target is irrelevant for the downstream one.