History of CherryPy
Remi Delon released the first version of CherryPy in late June 2002. This was the starting point of a successful Python web library. Remi is a French hacker who has trusted Python for being ultimately one of the greatest alternatives for web application development.
The project attracted a number of developers who were interested in the approach taken by Remi:
CherryPy classes were an extension to Python to support the separation of concern between the data and the presentation. It was close to the model-view-controller pattern.
A CherryPy class has to be processed and compiled by the CherryPy engine to produce a self-contained Python module embedding the complete application as well as its own built-in web server.
CherryPy would map a URL and its query string into a Python method call, for example: http://somehost.net/echo?message=hello would map to echo(message='hello').
During the following two years, the project was supported by the community and Remi released several improved versions.
In June 2004, a discussion started about the future of the project and whether it should continue with the same architecture. One of the main concerns was the compilation step, which did not feel natural to Python developers. Brainstorming and discussion by several project regulars then led to the concept of object-publishing engine and filters, which soon became a core ingredient of CherryPy 2.
Eventually, in October 2004, the first version of CherryPy 2 alpha was released as a proof of concept of these core ideas. Then followed six months of intense work to release a stable version (late April 2005). Soon other developers joined the project to improve it. CherryPy 2.0 was a real success; however, it was recognized that its design could still be improved, and needed refactoring.
After further community feedback/discussions, CherryPy's API was further modified to improve its elegance, leading to the release of CherryPy 2.1.0 in October 2005. This version was shipped by the popular TurboGears project—itself a stack of projects to produce a web mega-framework. The team released CherryPy 2.2.0 in April 2006.
CherryPy's presence as a core ingredient in the increasingly widely adopted TurboGears stack naturally meant that more and more issues were raised about some aspects of CherryPy. For example, its WSGI support, the lack of up-to-date documentation, or its only-average performance. It was clear that to meet these real and important requirements, it would be extremely difficult to extend CherryPy 2 without breaking backward-compatibility constraints. As a result, the decision was finally made to move towards CherryPy 3, which was released at the end of 2006.
