[swift-evolution] Fixing the confusion between non-mutating algorithms and single-pass sequences

[Jonathan Hull]

I have also been thinking about this problem for the last week or so (as well as the finite/infinite bit).  I don’t really have anything detailed that is ready to share (and it sounds like you are headed in a different direction now).  I still wanted to share the gist of my thoughts, in case they help spark ideas in others…

My thought was to follow the first rejected approach: removing sequence and letting the Iterator protocol model single-pass.  Iterators would be reference types.

I followed a similar path, and my version also has a pretty large duplication of API between Iterator and Collection… the difference though, is I think I have a way to avoid most external duplication of API.

Basically, I added back in a super-minimal protocol to fill the structural gap left by Sequence.  I call it “IteratorProvider” and it only has a single function which vends an iterator.  Collection adheres to this, and Iterator adheres to it by returning itself.  All of the other methods from Sequence remain on Iterator.  Thus anyone with API that only needs a single pass would take a IteratorProvider and then work on the iterator it provides.

The big difference is that Collection and Iterator are still separate protocols, iterator is a reference type, and most of the methods from sequence are now on iterator.

I think this makes more sense semantically than the current model (or renaming sequence).  I also really think it is important to have iterators be reference types (anything else is really a lie)


The rejected “consumedIn” idea also gave me an idea of how to reduce the internal API repetition, if desired. 

Have a fileprivate method on Iterator (I will call it “consumedIn” here, but it is private, so call it whatever) that wraps the Iterator in a collection. The multi-pass-ness of that secret collection is a lie, but it is fileprivate so it should never get into the wild where someone can find that out. Then you would just define map() etc… on an extension of Iterator and have them forward to “self.consumedIn.map”, etc….  It does still have duplication of definitions, but the implementations would be in a single spot.

Another option, if the subterfuge of a secret collection is undesirable, would be to make “consumedIn” be public and have it create an array-like collection.  The default implementation would actually make an eager copy, but specialized cases could work with the created collection to avoid copying the iterator’s contents where possible.  Then you would remove all of the eager methods from Iterator and just use collection’s version.

Food for thought…

Thanks,
Jon

> Hi,
> 
> I'd like to continue the discussion of the issue raised by David Waite
> inhttp://thread.gmane.org/gmane.comp.lang.swift.evolution/21295/: <inhttp://thread.gmane.org/gmane.comp.lang.swift.evolution/21295/:>
> 
> > My main motivation for proposing this is the potential for developer confusion. As stated during one of the previous threads on the naming of map, flatMap, filter, etc. methods on Sequence, Sequence has a naming requirement not typical of the rest of the Swift standard library in that many methods on Sequence may or may not be destructive. As such, naming methods for any extensions on Sequence is challenging as the names need to not imply immutability.
> 
> I'd like to focus on a particular point: methods on Sequence can
> consume elements, but the APIs are not markedmutating.
> 
> Dave Abrahams, Max Moiseev, and I have discussed this issue and we
> agree this problem is severe and worth solving, we also think that the
> likely solutions would be source-breaking, so it is important that we
> discuss it now.
> 
> We have discussed a few options.
> 
> - Rejected option: remove Sequence, let IteratorProtocol model
> single-pass data streams
> 
> - Rejected option: use a syntactic marker, like sequence.consumedIn.map {}
> 
> - Rejected option: mutating APIs on Sequence, non-mutating APIs on Collection
> 
> Proposed: rename Sequence to IterableOnce or TraversableOnce. We think
> that Sequence does not convey the single-pass restriction clearly. The
> term "sequence" has been used in math (as in "integer sequence"), and
> since the math domain does not have mutation, "sequence" can be
> understood to mean "multi-pass", since you can traverse a sequence of
> integers an arbitrary number of times.
> 
> We think that only the last option is viable in the Swift language as
> it exists now, without creating an undue burden for API vendors and
> users.
> 
> For more details about rejection options, please see the full writeup:
> https://gist.github.com/gribozavr/47f4717f3afc762549383e94da7f748b <https://gist.github.com/gribozavr/47f4717f3afc762549383e94da7f748b>
> 
> Dmitri

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.swift.org/pipermail/swift-evolution/attachments/20160714/7ab46d40/attachment.html>