Trouble with sparse matrix transpose in 0.7 (and now 1.0 !?)

I disagree, notational convenience should favour math-like operations. I also suspect this is the more common use-case. In your use-case you can achieve the wanted behaviour using collect, with zero performance loss.

In general, it is simpler to take a lazy operation and make it eager, than the other way around.

As for my example, it is trivial to come up with one where copying behaviour would have significant cost.

1 Like