I was working at Promitheia during the first days of “Web 2.0”. AJAX was very popular but the tools to fill the gap between dynamic and static sites were not yet around. Server side rendering was necessary, but I still wanted to provide very interactive pages without adding a completely different client-side framework for rendering markup.
I settled on what some called AHAH - Asynchronous HTML And HTTP - where the server rendered partial views via XMLHttpRequests which were inserted into the DOM. I made use of a number of non-standard HTML attributes to declaratively wire up these insertions. As a result very few widgets needed custom code. jQuery was patched at startup to rebind events when content was inserted. I was able to achieve performant and flashy interactions with careful use of jQuery (it is possible!) long before they became commonplace.
Working with views in ASP.NET prior to the release of Razor was painful: unmaintainable soup markup was everywhere. As a result, many people were creating their own view engines. I did not have the bandwidth to create and maintain a custom view engine, let alone make it work with the IDE.
I decided to create a C#-based fluent DSL and write the views directly in the host language itself. People now call this hyperscript, and it is fairly common, but at the time (2008-ish), other programmers thought I was insane! This had many benefits - I did not have two syntaxes mixed up in the same file, I had all the support of my IDE (autocomplete, type checking), and I could package components as DLL files. Layout views could simply be written as methods with parameters for the “holes”.
In debug mode my view engine would set up a file watcher and auto-reload views as I worked on them; another perk which was scoffed at then but is now looked at as a standard feature of a web stack.