to_the_sun

If I store a small amount of data (maybe a couple dozen items) in a coll, but at indices that are very large (perhaps in the millions) will it be equal in efficiency to the same scenario except stored at small indices?

I would like to improve my code by storing my data in a buffer~ or perhaps a [data] in gen~, but having to declare the size is then a problem.

using-coll-as-a-sparse-array

Without looking at the underlying code, I'm fairly certain that coll operates relatively efficiently regardless of the indices. I suspect it is internally implemented as a linked list.

But, I'm 100% certain that dict should be fine in this case, and is probably better than coll in almost all cases.

Without looking at the underlying code, I'm fairly certain that coll operates relatively efficiently regardless of the indices. I suspect it is internally implemented as a linked list. 

But, I'm 100% certain that dict should be fine in this case, and is probably better than coll in almost all cases.


The values used as indices should not effect CPU performance of a coll, so go ahead.

IMS, coll is just a linked list of entries. It's been years since I looked at this closely, so possibly double linked, or the links may be inserted to keep entries sorted in index order. But what stayed in my mind was that most performance issues were O(n) for number of items stored (including searching for an index); performance would not be effected if your indices are {1, 2, 3,…} or {19, 1483, 195467, …}.

Yes, thank you. Is there any way to utilize a sparse array like this in gen~ (or somehow in the signal domain otherwise)?

Using coll as a "sparse array"