> On Mar 19, 2017, at 2:21 PM, Tony Parker <[email protected]> wrote: > > >> On Mar 19, 2017, at 12:14 PM, Matthew Johnson <[email protected] >> <mailto:[email protected]>> wrote: >> >> >> >> Sent from my iPhone >> >> On Mar 19, 2017, at 10:47 AM, Tony Parker <[email protected] >> <mailto:[email protected]>> wrote: >> >>> Hi Matthew, Brent, >>> >>> I see why you are asking for this Context parameter, but putting it into >>> the basic Codable protocol introduces too much conceptual overhead. There >>> are too many benefits to keeping adoption to just one protocol, including >>> discoverability, ease of use, reducing the need for overloads on protocols >>> elsewhere, and more. Supporting this one use case does not outweigh those >>> benefits, especially considering I expect that most library code would not >>> use it (as you say: it would be weird to pass this context between modules). >>> >>> Can you figure out a way to get the context info passed through the >>> encoder/decoder instead? It would make more sense as something optionally >>> retrieved from the encoder/decoder that was set at the top level. >> >> Hi Tony. I can see the argument that the this is a feature that should be >> relatively rarely used and thus should have as simple a design as possible. >> >> If you feel like the impact of threading a typed context on the API surface >> area is too heavy you could just add a `var context: Any? { get }` >> requirement to Encoder and Decoder. The expectation is that encoders and >> decoders would accept a context in the top level call and make it available >> to all Codable types. This would solve the problem with minimal API impact >> at the cost of the ability to statically verify that all types receive the >> context they need to encode / decode correctly. >> >> I much prefer the static safety but having a solution is better than not >> having one. :) > > The Any context property is reasonable, but it would be nice to find > something in the middle. =) > > One other possibility is that we define a user info dictionary instead, with > a custom key type that can be extended (much like our string enumerations). > In general I haven’t been a fan of the user info pattern in Swift because of > the necessity to cast, but as you say it’s better than nothing. e.g. userInfo > : [CodingUserInfoKey: Any].
This makes sense some sense. This would allow us to support the multiple context use case. I think the need for that is far more rare than the need for a single context but it still makes sense to support it. The down side I can see is that it could encourage users to adopt a “context dictionary” approach with a bunch of keys rather than defining their own context type using a single key. This is generally a bad idea and should be discouraged. I think we should make the more common single-content use case more convenient and subtly nudge users in the right direction by making it easier to use. We could do this by defining a `DefaultContext` key, including a `defaultContext` property on `Encoder` and `Decoder` in an extension which returns `self.context[DefaultContext]` and encouraging encoders and decoders to provide an override that takes a single `context: Any` argument which gets placed in the context dictionary using that key. I still very much prefer the stronger guarantees offered by Brent’s first design but I could live with this. One last thought - might it be possible to restructure the design a way that would allow us to build type-safe context-awareness on top of existing encoders / decoders? I’m going to give a little bit of thought to this but don’t expect to come up with a good answer. At least one hurdle to this in the current proposal is that the Foundation encoders and decoders hide the types that actually do the encoding and decoding. There are probably very good reasons for this of course, but they also make it more difficult to layer functionality like context-awareness on top of them. We would have to resort to a hack of some kind to even attempt it. > > - Tony > >> >>> >>> - Tony >>> >>>> On Mar 17, 2017, at 5:56 PM, Matthew Johnson via swift-evolution >>>> <[email protected] <mailto:[email protected]>> wrote: >>>> >>>>> >>>>> On Mar 17, 2017, at 6:15 PM, Brent Royal-Gordon <[email protected] >>>>> <mailto:[email protected]>> wrote: >>>>> >>>>>> On Mar 17, 2017, at 3:35 PM, Matthew Johnson <[email protected] >>>>>> <mailto:[email protected]>> wrote: >>>>>> >>>>>>> In all seriousness, I see the design as very slightly weak, in that it >>>>>>> makes it easy to forget to pass a context through, but quite >>>>>>> acceptable. >>>>>> >>>>>> Easy for who? I was not requiring Codable types to thread it through at >>>>>> all. The context was fully managed by the Encoder / Decoder type. The >>>>>> only place Codable types work with the context is as an argument they >>>>>> receive. They never pass it when encoding or decoding anything. The >>>>>> Encoder / Decoder would need to store the context internally and when >>>>>> call is made to encode / decode a ContextAwareCodable it would pass the >>>>>> result of a dynamic cast to ContextAwareCodable.Context as the context. >>>>> >>>>> Oh, I see. Sorry, I missed that when I was looking at your design. >>>>> >>>>> In practice, in my design, you would only need to manually pass a context >>>>> to `encode(_:forKey:with:)` if the context was of a different type than >>>>> `self`’s. >>>> >>>> Oh, I see. I missed that part of your design. I really like it with the >>>> shorthands. I’m fully on board with this being the right way to handle >>>> contexts now. I think Context should be in the basic Codable protocol. >>>> That leaves the question of what to do with NSKeyedArchiver and >>>> NSKeyedUnarchiver. I’m not sure what the answer is for those but it would >>>> be unfortunate to see the design compromised solely because of a >>>> requirement to interoperate with them. >>>> >>>>> This would probably happen at module or subsystem boundaries. Imagine, >>>>> for instance, that your FooKit module (for interacting with the foo.io >>>>> <http://foo.io/> web service) needs to encode a GeoKit.Location instance, >>>>> but both FooKit and GeoKit need information from a context to encode >>>>> themselves properly, and they use different context types. When FooKit >>>>> encoded a GeoKit.Location, it could construct and pass a GeoKit context. >>>>> >>>>> I believe that in your design, unless the FooKit context was a subtype of >>>>> the GeoKit context, you wouldn't be able to get GeoKit.Location to do the >>>>> right thing. >>>> >>>> Right. It was assuming only one context would be needed for an entire >>>> encoding / decoding process. I don’t know of use cases where one module >>>> could meaningfully provide a context to another module unless they were >>>> very closely related (i.e. built as parts of the same system) but maybe >>>> they do exist. Your design is able to accommodate this very well. >>>> >>>> I made some compromises to try and diverge from the current proposal as >>>> little as possible while still solving the primary use cases I’m aware of. >>>> Now that I understand your design I think it has enough advantages that >>>> we should go in that direction. And we certainly should not go in the >>>> direction of something that requires Any. >>>> >>>>> >>>>> If that weren't the case—if you were encoding a type with a matching >>>>> context, or with a `Void` context—you could use the two convenience >>>>> methods, which would handle the context argument for you. So threading >>>>> contexts would only be necessary in a relatively rare case. >>>> >>>> Yep, that’s very elegant! >>>> >>>>> >>>>> -- >>>>> Brent Royal-Gordon >>>>> Architechies >>>>> >>>> >>>> _______________________________________________ >>>> swift-evolution mailing list >>>> [email protected] <mailto:[email protected]> >>>> https://lists.swift.org/mailman/listinfo/swift-evolution >>>> <https://lists.swift.org/mailman/listinfo/swift-evolution> >
_______________________________________________ swift-evolution mailing list [email protected] https://lists.swift.org/mailman/listinfo/swift-evolution
