Cioccolata

FastCGI web application framework in Objective-C

d11wtq
3
0
C

Cioccolata - A FastCGI embedded framework for creating web applications in Objective-C

This project is as yet barely even born, but the intention is to create a simple web-
application container based around Cocoa, capable of running either on a Mac, on Linux
via GNUstep or Cocotron, or on Windows via Cocotron. Initially it will be built to run
on Mac OS X 10.5 or later, then gradually adapted to run under GNUstep.

I’ve always questioned why C-based web environments do not exist, and the response
always seems to be “they’re just not very web like”. Well, as an experiment I’d like
to see just how much of my web-development experience I can bring to the objective-C
environment. I believe that Objective-C would be a welcome addition to the
web-developer’s toolbox, offering a great deal of functionality, including of course,
the potential to bridge with interpreted languages such as Ruby, PHP and Python if
wanted/needed.

With the advent of HipHop (Facebook’s PHP to C compiler) it’s clear that people are
looking for more efficient new ways to run applications on the web. Cioccolata could
grow to fill that void. It needs to be simple to use, but not constraining with
“Cioccolata-ways-to-do-things”.

Initial goals (next few months):

  • Create the controller subsystem in the most flexible way possible
    • Ruby-like routing
    • Pre-dispatch hooks
    • Action controllers / ViewControllers
  • Create a template engine inspired by django and others held in high regard
    • Template inheritance
    • Customizable Filters
    • Customizable template logic (tags)
    • Clear promotion of separation of concerns

Up-coming goals:

  • Multiple-DBMS abstraction layer (Active Record or Datamapper?)
    • Table inheritance
    • Object-persistence
  • Create a solid data-input and validation framework
    • I’m still yet to see one that I truly like but can borrow bits and pieces from others.

Once the base framework (each of the M-V-C components is in place, I will build something
of reasonable scale, designed to support being run in a cloud, using this as a way to
guage feature additions and needed refactors.

Biggest issues forseen?

  • Creating the multi-DBMS framework in an abstract manner will be very time-consuming
    • Will keep looking for existing solutions
    • It should be two layers… one for the DB-access, the other for ORM.