chore: improve errors in iterators.R

[maelle]

Fix #1891
Part of #731

[maelle]

I wonder whether we could improve this.

[krlmlr]

Let's split this PR, the "Graph is unknown" thing is special and might become unnecessary with the new attribute handling, I hope.

[maelle]

Sorry, but split into what? One PR with all error messages minus the one I commented on?

[maelle]

#2006 (comment)

[krlmlr]

Let's find or generate extensive tests for the intended behavior of this code. The RHS for these assignments is an object with name and value attributes, I'd like to understand actual usage examples.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 2be846a is merged into main:

• ❗🐌as_adjacency_matrix: 744ms -> 765ms [+2.04%, +3.58%]
• ✔️as_biadjacency_matrix: 743ms -> 741ms [-0.83%, +0.27%]
• 🚀as_data_frame_both: 1.64ms -> 1.55ms [-8%, -2.78%]
• 🚀as_long_data_frame: 4.02ms -> 3.91ms [-3.83%, -1.55%]
• ✔️es_attr_filter: 2.68ms -> 2.73ms [-0.34%, +3.55%]
• ✔️graph_from_adjacency_matrix: 118ms -> 118ms [-1.11%, +1.38%]
• 🚀graph_from_data_frame: 3.73ms -> 3.46ms [-8.76%, -5.96%]
• ✔️vs_attr_filter: 1.54ms -> 1.52ms [-3.75%, +1.33%]
• ✔️vs_by_name: 980µs -> 979µs [-1.71%, +1.57%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[krlmlr]

I wonder if we should just make <- NULL work idiomatically -- drop attribute. We could assess the revdep impact.

[maelle]

I had added the hackathon label to #1891 😸

[krlmlr]

Can we solve #1891 as part of this PR?

[maelle]

I asked Claude to rename name and value to attr_name and attr_value in this context because I think the double "value" added to my confusion when I first looked at this code.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if f46e70f is merged into main:

• ❗🐌as_adjacency_matrix: 872ms -> 926ms [+3.3%, +9.23%]
• ✔️as_biadjacency_matrix: 871ms -> 861ms [-2.93%, +0.56%]
• 🚀as_data_frame_both: 1.85ms -> 1.74ms [-8.96%, -2.95%]
• ✔️as_long_data_frame: 4.41ms -> 4.37ms [-5.35%, +3.31%]
• ✔️es_attr_filter: 3.35ms -> 3.35ms [-2.81%, +2.45%]
• ✔️graph_from_adjacency_matrix: 141ms -> 143ms [-1.92%, +4.74%]
• 🚀graph_from_data_frame: 4.06ms -> 3.82ms [-9.25%, -2.43%]
• ✔️vs_attr_filter: 1.85ms -> 1.86ms [-1.87%, +3.09%]
• ✔️vs_by_name: 1.23ms -> 1.21ms [-4.3%, +1.14%]
  Further explanation regarding interpretation and methodology can be found in the documentation.