Re: Does anyone want intersperse for Data.Sequence?

It depends what you mean. Seq has nice asymptotics but quite poor constant factors. Do you consider DList as requiring a separate building phase? If so, the answer is quite possibly that the data structure you want doesn't exist. But for most code I think the existing solutions are good enough (and performance requirements lax enough) that nobody has put much effort into anything better. John L. On 08:30, Sun, Dec 28, 2014 Greg Weber wrote:

I want an interface that supports efficient appending without requiring a separate building phase or other hassles. Does this not exist?

I get that a library author might want to deal with the hassle of alternatives. But as an application author there are times that all I care about is a clean and simple interface that supports efficient appends.

On Sun, Dec 28, 2014 at 1:01 AM, Roman Cheplyaka wrote:

...

On 27/12/14 17:03, Greg Weber wrote:

...

Sequence is not used nearly enough.

Not really. Seq has high constant factors and for small sizes is often slower than plain lists. Nor is it space-efficient (for big sizes).

It probably has its use cases, but most of the time you'd be better off with one of the other choices:

* plain old lists * difference lists (trees) * lists "without remorse" * vectors * lazy vectors (lists of vectors)

Roman

_______________________________________________ Libraries mailing list Libraries@haskell.org http://www.haskell.org/mailman/listinfo/libraries