about making magritte more flexible... and founding the case for L17N

Hi, This is a "disgression" of something I thought/implement last week and I want to put into consideration, in order to know if is a valuable contribution or just another stupid idea :) First let mu put some context information: I'm using magritte for some production applications, who has two major requirements: L&F and L17N. For major of my needs, magritte fits quite well, but it still have some restrictions: 1) No L17N in labels and buttons. 2) Buttons are texts and usually clients ask for images, images/texts, etc. There are some other restrictions, but achievable by simply extending magritte system (such as image report columns, etc.) How do I solved this problems? Well, first I simply added a "localized" method on String who asks for a property (language) in session and follow "translated" path to translate the string, and then I modified each method in magritte who outputs a label (they are not many, just 3 or 4). That solves my L17N problem. Now, I want to make my buttons more fashionable... this is harder because of the places where I have to change things... but again, extending MAFormDecoration and MACommandColumn... and aplying that to every component who uses a report... problem solved. But, what if I what my buttons different, or even configurable? This is my approach: instead having buttons and links displayed in columns or forms, I created a Command, and then I whant to add that commands by visiting different renderers, magritte style visitors. So, if I want to change the displaying style, I just add a new renderer... And as an "extra bonus" there I can handle L17N stuff. What do you think? is this good approach?