NamedRows::next documentation

[dcz-self]

Originally I thought NamedRows::next was a kind of a cursor structure, to avoid storing large results all at once in memory, but after taking a closer look, the chain stores all data together anyway.

The instance in next can also store its own columns. Is this for reporting results from multiple queries?

Could the documentation explain when the user can expect next to be populated, and whether the next set of rows is actually independent from the first?

[zh217]

It is only used for returning multiple relations within an imperative script, which may also be undocumented (I can't remember).

I will remind myself to document these when the next release comes.

[dcz-self]

I mention this because if the columns can change between instances, then my latest pull request is buggy.

[zh217]

Thank you for pointing this out. I didn't look at that carefully enough. Can you make an update pull request?

[dcz-self]

If columns can be different, then the entire change must be reverted. I don't know how to improve that API without knowing more about the next field.

[zh217]

I see. I will revert it.