On Sun, Sep 27, 2009 at 9:41 PM, Zarel <zare...@gmail.com> wrote: > What is so wrong with maintaining two drawing paths? We should be > modularizing our codebase, anyway, so that everything that touches > OpenGL directly should be in a few library files. After that, we can > just leave the legacy OpenGL code alone.
It simply does not work like that. The internal data format changes require to accommodate more modern ways of drawing graphics, or simply changes to introduce new graphics features, will necessitate changes in all the graphics code, old as well as new. You cannot just convert the data on the fly, because that would be too CPU consuming, and you absolutely cannot maintain two internal data formats by the way of massive #ifdefs in the code. This was a painful learning experience when we tried to maintain the software renderer. In the end we had to choose between improving the OpenGL code path, or dropping the software renderer. We did not have the manpower to do both. > I'm sorry, but I can't really support anything that would drop support > for the majority of our players. We are not Oblivion, or Crysis, or > SupCom; we do not have the marketing department, good graphics, and > general stability/reliability necessary for us to survive losing the > vast majority of the market segment. There is a balance between keeping the hardware demands down on the one hand, and improving the graphics and giving the artists what they want on the other hand. By keeping the archaic data formats required by the old rendering engine and its lack of features, we lose interest from artists. As long as we do not get to improve graphics to match what is expected by players today, we will not attract as many new players as we could have. So while keeping hardware requirements down can make us not lose existing players and people who played the original ten years ago, ramping up the hardware requirements and adding better graphics *could* win us many times more new players (with gamer type hardware). However, since we are not a commercial entity, we have no need to make decisions like this based on perceived market share gains. Basically, we have to find a good place to draw the line. I have learned now that we already use VBOs in trunk for the new terrain drawing, which would break my Intel GMA X3000 on Linux back when I had that. Maybe the drivers have improved since. The X3000 has the hardware to run fragment shaders in theory, but it did not work when I tested it a year ago. Maybe it works now. To make a good decisions we need facts like that. - Per _______________________________________________ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev