Saturday, June 23, 2012

What's Next

My Thesis included the "release" of Abundant 0.3, which was somewhat minimal, but usable. In my thesis, I identified several specific missing or incomplete features that would be necessary for a 1.0 release. I'll enumerate them here, and try to create a blog post (or several) as I resolve them.

Each of these goals correlates to an increment of the minor version number, and accomplishing all of these goals would effectively indicate Abundant is ready for a wide scale 1.0 release. You can follow my progress via the 1.0 Track tag on this blog.

Read-only web interface
We should replicate the functionality of Mercurial to serve up information on the Abundant database in a browser, either from a central server or from the local computer. This should be painless to do, and allow the user to quickly browse the current database state.

Interactive web interface
Abundant is crippled until users can update issues without needing to first pull the source code down and then push it back up. As such, we need to expand the web interface to allow users to submit new issues.

Speed improvements
There is a notable startup delay at present, which is tolerable for now, but absolutely cannot continue. Standard commands, like init, new, and details must execute nearly instantly. More complex commands like list/tasks can be slower, but still cannot be as susceptible to slowdown as they are now.

Advanced workflows
Currently Abundant is fairly free form in its workflow, issues can be created, changed, and closed by anyone. This is desirable for small projects, and remains a good default behavior,however larger projects need greater control. For instance, it might be necessary for the person filling an issue to be the only one able to mark it resolved (user A files and assigns to user B, user B implements fix, marks it as fixed, it gets reassigned to user A, user A checks that it is resolved, closes it). These more powerful workflows need to be as simple to set up as possible, and these design decisions have not been made yet.

Data Caching
As it is, Abundant does not cache any information, and has to poll each individual file in order to get necessary information, including something as simple as the issue’s title. We need to implement caching which is transparent and failsafe (the user doesn’t need to know it exists,and when it’s outdated or broken, the system still runs, albeit slower) so that searches and lookups can be run faster. Examples include getting titles of related issues without having to load whole issues, and constructing sorted data structures to filter the list command faster.

Unit and Regression Testing
There may be some rudimentary testing being done before graduation, however it will be important if this project is to grow that a standardized and stable testing suite is developed to ensure the codebase does what is expected, and that future changes do not cause regressions that break previous functionality.

Improved User Interface
Notable when viewing the details of an issue, currently it is output in a format that is difficult to read, and in an order that isn’t necessarily desirable. We want to ensure the look and feel of the command line is professional.

Installation Scripts
It should be trivial, especially on Windows / Mac to install Abundant. Currently it isn’t terribly painful – pull code, add command to path, run – but isn’t the sort of installation flow users of these operating systems expect. It is ok for now to ask users to go through those steps,but for a large scale release to be possible, we must have installation binaries, which are preferably built by an automatic build script.

No comments:

Post a Comment