Rename genrows()?

[bluss]

.genrows() has an unfortunate name. It's supposedly the generalized rows of an array. The reason for the tricky name, was that .rows() was taken - by the row count.

We could rename rows →row_count and genrows → rows to solve this, ideally with deprecations, so in a two version cycle.

rows() is a more natural name for the ndproducer and iterable for rows, for example because the user doesn't have to care that it's “generalized” rows if they only use 2D arrays — they don't need to know and it doesn't matter.

The main problem with the genrows name is that it is non-intuitive to guess to type and guess what it means.

Note that we chose the .rows() name for the row count to match the sprs crate, but I think we can choose to go our own way here, if we have a good reason.

An alternative to the row_count name is nrows() — anyone have a survey of what other crates prefer apart from sprs?

NOTE: The same story applies to gencolumns.

[jturner314]

I agree that .genrows() has an unfortunate name. It's funny, I've always read it as "generate rows" instead of "generalized rows", which confused me because none of the other iterator/producer methods had the gen prefix. .rows() does seem like a better name.

Other crates:

• nalgebra: nrows and ncols are the counts. (rows and columns are slicing operations. row_iter and column_iter are the iterators.)

• rulinalg: rows and cols are the counts. (row_iter and col_iter are the iterators.)

• sprs: rows and cols are the counts.

My preference for the counts is nrows and ncols/ncolumns because these names are clear while being shorter than *_count.

[LukeMathWalker]

I agree that .genrows() has an unfortunate name. It's funny, I've always read it as "generate rows" instead of "generalized rows"

Same here 😅

My preference for the counts is nrows and ncols/ncolumns because these names are clear while being shorter than *_count.

👍