bwhiteley / jsonshootout Goto Github PK

I noticed the test target runs without whole module optimization enabled (since it runs the test in debug mode). I wonder if it would be a good thing to include? Though not every developer builds with it enabled.

The current JSON sample uses Strings for every kind of value, for a more complete view, it may be worthwhile adding another JSON sample that covers more of what you would expect to see served from a REST API. Perhaps something like https://raw.githubusercontent.com/vdka/JSON/master/Tests/JSONTests/Fixtures/large.json.

It would be great if you could include some test results for different sized JSON files. Mapping large JSON files is a very specific use case which I would argue is not that common. Most APIs typically will serve paginated data with 10-50 root level objects.

I think it would be useful to see how the parsers perform with varying sized data sets.

Full disclosure, I am the primary developer of ObjectMapper.

Great comparison but you miss good reflection based object mapping library: https://github.com/evermeer/EVReflection

I see it in Pull requests.

Before I pull and run the tests myself, in the article you state:

"This graph shows time spent in each of the mappers as well as time spent in NSJSONSerialization for a reference."

However, I do not see this data point anywhere in the graph... Please help me if I'm too near-sighted to see it.

This comparaison are good, but you miss for me the better solution for converting JSON <-> Data Model in two way

You should add this solution to your test case : https://github.com/Hearst-DD/ObjectMapper

There's a second pure-swift JSON parser which has been around for a while, which is my very own PMJSON. I'd love to see it included in the shootout.

Any chance we could get Genome on here? I'm curious how it stacks up. 👍

https://github.com/matthewcheok/JSONCodable

https://github.com/muukii/JAYSON

Any plans to update this now that Xcode 9/Swift 4 are shipping? Curious to see if anything has changed.

How about adding this library, which is used often as well (900+ stars)

https://github.com/delba/JASON

https://github.com/Anviking/Decodable

https://github.com/bignerdranch/Freddy

I may be misreading here, but in your section “That Thing on the Swift Blog,” you reference a Swift blog post from 2016. The concluding section (“Reflecting on Reflection”) of that post argues not against third-party frameworks, but against using reflection.

However, there’s a newer Swift blog post (well, I could have sworn I saw an official Swift blog post) that uses Swift’s JSONDecoder, which uses reflection by default. I thought that article argued against third-party libraries, but now I just can't find it (perhaps it was not an Apple or Swift post. Here’s one). I tend to use Marshal, but I’m all for removing dependencies. I'll probably switch to JSONDecoder(), but I was curious about its performance. it’s probably different when using reflection vs. explicit decoding.

Hey there, thought you might be interested in this:

All the performance difference between Marshal and Mapper is because of this one commit: lyft/mapper@84b7fb4.

Basically they use their own value(forKeyPath:) function so they don't get exceptions for type mismatches.

I'm not saying this is good or bad, just interesting.

https://github.com/hkellaway/Gloss

What the title says. Made by thoughtbot. Find here: Argo

What versions of Unbox, Marshall, and Argo were used in the analysis?