phrostie said: (by the date of Sun, 7 Sep 2003 21:10:40 -0400) > in the basic case it would be simple. the basic views(plan, top, bottom, > left and right) would be created by taking one of the major axis and reducing > it to zero. then converting it to something like dxf. where multipule views > can be detailed and demensioned. as time goes on more advanced features > would need to be added. the ability to render isometric views, auxilery > views, section views, section cuts, and filtering man! that is _great_ idea. I understand the idea *one* program does *one* thing. But my imagination is not enough to imagine how things can get separated. I thought of a "display"-server that renders (displays) the edited file, then allows the user to select objects, and click to enter coordinates. > the advange is that it would not matter who's modeler or who's drafting > package we use. as long as they work within the unix philosiphy. > "each tool does one thing" and they all work together. The main concern would be then that this .dxf renderer is fast enough to allow interactive display of what we are working on. Suppose this renderer uses our filesystem-based data format. Then it could pipe it's dxf output directly to "display"-server - which then displays it. Each program can be improved separately. And this way - we can put a very complicated task of reading data (from the filesystem based data format) in a one program. That does only this one thing. > i'm rather determined that dxf would be a roseta stone. even though we may > all want a common open format, we still have a real world we have to live > within. just about everyone from catia to unigraphics can read a dxf. I agree. .dxf should be good just to represent a 2D data just to be displayed on the screen while working on vector drawing. As input you give parameters of viewport. As output - you have a small (it should be small, shouldn't it?) .dxf. -- # Janek Kozicki