Wednesday, April 8, 2009

What Trac really needs

Anyone who has read this blog before might have found my previous post where I mentioned how I started getting into setting up my own web server on and that I was also installing Trac, which according to the Trac documentation:
"Trac is a minimalistic approach to web-based management of software projects. Its goal is to simplify effective tracking and handling of software issues, enhancements and overall progress."
Well, I have got Trac up and running. Trac actually relies quite heavily on its command-line client which, to be honest, I have no problem in using. Anyone that develops and/or works on and for a *nix environment is probably more than comfortable using the command-line and probably, like me, finds it far more useful and efficient than any GUI could probably be. There were a few issues however in setting up Trac that I thought I would share here for anyone reading and interested in setting up their own Trac installation.

1. Command Line requires a learning curve

This may seem counter to what I said above but the one advantage a GUI interface has over a command-line is that it is intuitive. With a GUI you can see buttons and prompts beckoning you to use them. With a command line you need to know the commands or ... well you can do nothing. This means that anyone looking to install and run Trac as of now will have to spend extra time learning the, albeit rather basic, commands.

This is alleviated somewhat with a very useful help system as well as fantastic online documentation for Trac, but the fact still remains that that learning curve might put people off.

2. Root access needed

Trac is not a simplistic web application, even though the documentation calls it minimalistic. It requires the person installing to have root access to the machine and is one of the reasons why I am moving to running my own server as opposed to continuing with a shared, managed service as I have done in the past. While it is understandable to some degree because of the SVN integration, again, this requirement will limit the available user-base to those who know how to setup and maintain web servers or have enough dosh to fling around to get their server management company to install it for them.

3. No built-in authentication

Thats right. Unfortuantely Trac does not include its own authentication system, so managing multiple projects for different clients who should not have access to one anothers projects can be a little nightmarish. If you want authentication then Trac expects you to use Apache's own built-in authentication system's (or whichever web server you happen to have installed). This means that anyone installing this also needs to know how to setup Apache in order to authenticate users based on encrypted, password files stored on the server itself and referenced to using Virtual Host settings.

Again, this limits the potential users of Trac to those that are sys-admins or have the money lying around to get someone to do it for them.

4. Lets give Trac a break

I mentioned a few issues I had but lets cut the developers some slack. Why? Well, Trac is only at release 0.11. Yup! They haven't even reached a full release version yet and pretty much what is available is beta-ish. Once you know that Trac development is still steaming ahead and that the "issues" I described above will probably have solutions to make things easier once the development team do incremement that version counter to 1.0, it doesn't seem such a bad deal for a free development management and bug-tracking application. I am pretty sure that in the next few weeks to months we will see Trac become a feature complete system and I cannot wait for that day. So far I am very impressed with what I can do with it and am so glad I stumbled across it a few months ago.

No comments:

Post a Comment