Tell

Besides that Wilhelm Tell is the big liberty hero in the myths of the canton of Uri in Switzerland, Tell is a URI builder library written in Kotlin. It's implementation builds upon the relevant RFCs for URIs, which are:

• RFC 3986 - Uniform Resource Identifier (URI): Generic Syntax in progress
• RFC 8141 - Uniform Resource Names (URNs) TODO

Why would one need this lib? Are'nt Strings enough or even of better performance?

The driving ideas to this library are

• to have a smart DSL to assemble URIs with the safety of a compiler and thus to get rid of primitive obsession while building and passing around URIs via Strings, which is error prone and hard to debug
• to allow fine grained validation of each part of an URI
• to allow you to extend the DSL with your own types, which gives you a lot of expressiveness and power for the routing inside your application
• to have classifiers for important parts of the URI, which can be used by aspects, if needed

It's a fact, that an URI, although seemingly short to our perception, is one of the quite complex syntactic constructs in the context of the web. Web applications working with String primitives tend to produce maintenance nightmares. It can be bloody mess to debug, how a malformed URL was built, especially if it is being passed around and modified between different software modules or even multiple services in a micro service landscape. Furthermore it happend several times in projects I've been part of, that a given subset of URIs were subject to changed requirements - either regarding form or parameters or whatever. That's a common case, where a development team faces a shotgun surgery throughout the code base to modify all these URI occurrences of the given subset - to add a query parameter or a path prefix for example. Common hacks to fix this mess produce servlet filters, rewriting all matching links before the server response is committed to the client. This is a real performance killer, since the whole HTML document needs to be processed with according regular expression matches and even be modified. This type of modification is hard to test and hard to control in real life production scenarios. If instead this subset of URIs is classified and thus can be accessed by an aspect, such modifications become very simple.

Ah! And by the way: Please forget about that performance nonsense... (Not convinced? Have a look at this!)

A real world example...

TBD

Your own DSL? Dead simple!

TBD

Add tracking callbacks to a subset of your links (...the aspect thingy...)

TBD