How should I develop an unusual user interface?

Question

Say I have an idea for a desktop application with unusual interface requirements. I need:

• Pixel-level control over the look and feel.

• Finely specified custom controls.

• Platform-independence.

My typical solution in this scenario would be to use SDL (or SDL+OpenGL) as a graphics framework upon which to build the interface from scratch.

• Is it a good idea to work this way, or should I use an existing toolkit?

• If the latter, which toolkit with bindings for either C or C++ offers the cleanest and most versatile facilities for extension (creating new controls) and customisation (styling existing controls)?

Answer 1

If you actually have the graphics and design skills/resources, and the ambition, by all means do it yourself. Thats the only way you can, as my friends in the corporate would would say, 'realize your vision'. Or better, make your own re-usable GUI framework that you can use in other applications. Dont be dissuaded by mantras about re-use if you really want to do it yourself.

On the other hand, if you dont have those resources, either use a 3rd party toolkit, or re-consider the need for an 'outside of the box' interface. You'd be doing your users a favor. The only thing worse than a bad interface is a bad, ugly, home-grown interface.

I dont know what kind of apps you are making. But if we are talking normal windows/desktop apps, evaluate whether or not your 'outside of the box' interface is something thats actually good for the user. I've seen many apps with atrocious interfaces over the years because their developers didnt want or couldnt care to create an application that more or less followed interface norms. That creates a lot of unnecessary overhead on the user as they try to learn your product. And the thing is, their developers dont realize their interfaces are that bad. So unless you are working on the type of app that calls for something 'slick' (like a media player or game), if you dont have the resources do your users a favor and use a 3rd party or standard (for the particular platform) interface.

Answer 2

When should you work from scratch, and when should you use an existing toolkit?

Never work from scratch.

Always use an existing toolkit.

What toolkit offers the simplest, cleanest facility for extension and customisation?

You're already using OpenGL. What's wrong with that? Many, many things are built in it.

And all toolkits contain lower and lower levels of componentry that you can extend or modify. You have the source.

http://www.atai.org/guitool/

Answer 3

1. Only work from scratch if no other tool kits exist for what you are working on and if you think that is the case, double check to make sure it really is the case before reinventing the wheel. Also, more often than not, custom tool kits are one of the areas where you are going to be spending a lot of time working on code before writing a single line of code for your application. If you have a sizable code base and you think that it could be turned into a tool kit it might be worth investigating as a side project, but I wouldn't put it in the way of production projects you are working on.

2. The best tool-kit for extensions tends to depend upon what platform you are working on, know that that jQuery is really flexible is not much help if you aren't working on web applications.