* Is this being used in any apps in the App Store currently by you? (or Panic?)
* Is Alamofire a non-profit or some kind of organization? Curious why it's listed in the README.
* Perhaps a bit deeper: Why a rewrite of AFNetworking's feature set instead of a AFN 3.0?
Was there any reason why you chose callbacks over a Promise-like pattern for the async responses?
Also do you know of any good HTTP Server libraries that will let you develop simple HTTP JSON services in Swift?
Deeply-nested networking callbacks are a code smell, and I find that promises often do more harm by hiding those smells—or worse, obscuring the flow itself. In practice, Alamofire is mostly going to be single or at most nested two levels deep anyway.
Thanks for your work by the way.
Alamofire.request(.GET, "http://httpbin.org/get")
.header("custom-header", "value")The long answer: Custom headers on a per-request basis are a bit of a red herring, from an API design perspective. In reality, most use cases boil down to one of three, which are taken care of in other ways:
- Default headers, such as `User-Agent`, are specified in `NSURLSessionConfiguration -HTTPAdditionalHeaders`. - `Authorization` headers, which are handled by `NSURLCredential` and `NSURLAuthenticationChallenge`. - `Content-Type` headers, which are handled by `ParameterEncoding`.
This is why Alamofire is designed the way it is.
However, I think the library name is a bit of a poor choice. I don't want to take away your thunder, I just picked this post to do my rant :).
There seems to be a lot of new general-purpose libraries with very poor non-descriptive names coming out.
Whenever I need to dive into a new codebase, in a language and ecosystem I'm unfamiliar with, it's extremely hard to figure out which dependency and library does what. I was tasked to dive into a massive RoR application recently, and I ended up having to write a glossary first to remember what every non-descriptive noun implied.
After getting familiar with a particular language's ecosystem, this problem becomes tends to go away, and as a library writer, it becomes easier to fall into the same trap by picking a name that's marketable and distinct.
But I feel we're collectively making it increasingly harder for people to ramp up in new ecosystems and maintain future legacy codebases.
For names of large applications this is not as bad, and some of these applications become household names, but I think we should try to avoid this for general-purpose utilities.
but I, at least subconsciously, when faced with a library choice, take into account things like:
- Simple, pretty name (Alamofire is very nice-sounding and memorable – even feels like overkill, author should've saved it for an iPhone app or something.)
- The pretty logo
- The perfect use of Markdown (I'm so obsessed about using `` around everything that looks like code it feels like a psychological condition)
- Clear structure of README
- Simple examples with syntax highlighting
There must be correlation between the presentation and quality of the source code!
There isn't.
The overlap is relatively small between:
- People who are great at writing software.
- People who are great at designing logos and quality of presentation (or have someone on staff that is).
For what it's worth, I do love how the name sounds :)
This can go as absurd as the package named "a" in npm (https://www.npmjs.org/package/a) which is used by a lot other packages. I also had some trouble to figure out what each package is suppose to do when using RoR.
However, I would not say it's necessary easier in older languages when I see the amount of weird acronyms used in the Java world.
So... what do those guys at Panic have you working on? ;)