edwardt / scrumjet

RESTful task tracking experiment built with webmachine

ScrumJet: RESTful task tracking experiment built with webmachine

ScrumJet's experimental premise is simple: I wanted to learn more about [webmachine] 1. Basic task tracking is a fairly simple thing to implement and has lots of obvious ways to link things together.

Running the Experiment

Pre-requisitites:

• Mercurial (to get webmachine)
• Erlang/OTP

Shell 1:

make
./start-dev.sh

Shell 2:

./load-test-data.sh

Open [http://127.0.0.1:8000/] 8 in a web browser.

Motivation

REST(Representational State Transfer) is an architectural style which can be applied well to such a system and webmachine makes it easy to implement in [Erlang] 2.

Overall Architecture

ScrumJet is broken out into resources for tasks, categories, boards and utility resources for representing linkages amongst them. Much of the motivation for this architecture came from reading [O'Reilly's book RESTful Web Services] 3. As such it focuses on providing a well-connected HTML representation of each resource. I am likely to eventually add [JSON] 7 representations to the resources for efficiency, but my goal at the moment is a proper RESTful architecture. I find it easier to achieve that in HTML.

Part of the motivation for this architecture is a secondary educational goal of learning how to build an ARIA on top of a RESTful web service using [Dojo] 5. Obviously [Dojo] 5 prefers [JSON] 7 so some ScrumJet resources are also able to represent themselves as [JSON] 7. It is an end-goal that all resources in ScrumJet will be able to do this.

Underlying Architecture

ScrumJet's resources are implemented as [webmachine] 1 resources. Each resource generates an HTML representation based on state stored in one or more [mnesia] 6 tables. Specifically, each resource uses a gen_server to work with its primary state in an mnesia table and uses the gen_servers of other resources to determine connections or links to be rendered in the representation.