Repository navigation
Override Iterator methods when possible in std #24214
Description
Now that Iterator has re-emerged as a single trait, it's possible to override default methods like count, last and nth for iterators on slices, vectors, and so on. We should audit the applicable iterators and take advantage of overrides when possible.
libcore:
- slice::Iter(Mut): count, last, nth
- slice::Windows: count, last, nth O(1) count,nth,last for slice::{Windows,Chunks,ChunksMut} #27652
- slice::Chunks(Mut): count, last, nth O(1) count,nth,last for slice::{Windows,Chunks,ChunksMut} #27652
- char::{EscapeDefault, EscapeUnicode}: count, last, nth (techically but probably not worth it)
- str::Bytes: count, last, nth
- iter::Chain: count, last, nth
- iter::Enumerate: count, nth
- iter::Peekable: count, last, nth
- iter::Skip: count, last, next (should call nth), nth
- iter::Take: nth
- iter::Fuse: count, last, nth
Activity
aturon commentedon Apr 8, 2015
I'd be happy to mentor work on this bug.
djmally commentedon Apr 9, 2015
I would be interested in working on this.
GuillaumeGomez commentedon Apr 9, 2015
Me too.
tamird commentedon Apr 17, 2015
Also interested in working on this. Anything left to do?
djmally commentedon Apr 20, 2015
@aturon Any word on this?
pnkfelix commentedon Apr 20, 2015
Also, I'm curious if methods like
skip(n)are required to actually callnextn tines, or if they are allowed to "jump ahead" on random-access iterators. (Maybe that sort of trick will have to wait for specialization in the language, to avoid composing that with e.g. amapor other potentially side-effectful iteratorStebalien commentedon Apr 21, 2015
@pnkfelix
skipcan callnthon the underlying iterator and the underlying iterator can optimizenthif possible.pnkfelix commentedon Apr 21, 2015
@Stebalien I think that is my point: You say "the underlying iterator can optimize
nthif possible", but I claim to allow that would require changing the specification ofnth, which currently is specified as "going throughniterations", which I interpret as "calls thenext()methodntimes" -- a side-effectful iterator chain (e.g. one usinginspect) can observe the difference between that or jumping aheadnsteps when given a random-access iterator.aturon commentedon Apr 21, 2015
Oh dear, I seem to've lost track of this thread, sorry to everyone who was waiting on me to organize things!
I think probably the first place to start is just to get a list of the public iterators in std, which we can turn into a checklist and track assignees for. Part of the work will just be figuring out which cases are actually opportunities for doing better.
Anyone interested in putting such a list together? I'd be glad to help coordinate.
aturon commentedon Apr 21, 2015
(And if no one is in a position to make a list, I can put one together tomorrow.)
Stebalien commentedon Apr 21, 2015
@pnkfelix It doesn't because:
nth.slice::Iter) have side-effect-freenextmethods. These iterators can implement optimizednthfunctions.So
vec![1, 2, 3, 4, 5].iter().skip(2)can have a constant time skip given an optimizedslice::Iter::nthmethod. Unfortunately, as implemented, ranges can't optimizenththis way because they are generic.37 remaining items
Implement `EscapeUnicode::nth`
Implement `last` for `EscapeUnicode`
Auto merge of #33103 - ranma42:escape-unicode-last, r=alexcrichton
Implement `EscapeUnicode::nth`
brson commentedon Jun 27, 2016
From comments it appears this is done. Thanks all.