Hi boys,
I wouldn't like to move the flame forward, but I just can't keep it to myself.
Quote from: osw on 2011.05.18, 12:22:18
So give me one reason to vote for gradd.
The solution proposed by Demetrious - 3d accelerated driver for os/2 - may give us some cool possibilities - 3d api, compositing manager, alpha compostiting (you know - even smartphones have it now) - so new look and performance to os/2.
So, again - give me one reason to bother with gradd? Nostalgia? Dos sessions? Win-OS/2 sessions? (rotfl)
Please note that I'm following GL/2 with great hopes, and I'm absolutely not against its chosen directions, I just simply don't understand why would one want to throw out GRADD.
GRADD is a general Graphics Adapter Device Driver architecture, which was designed with an exceptional extensibility in mind, meaning that it was (it is) prepared for every existing and future feature of a graphics card. Even more, one can create filters to modify how an old GRADD device driver works.
Adding 3D capabilities to an existing OS/2 system which uses a GRADD-based video driver would just mean one more GRADD module/driver to be installed. So, a 3D driver would nicely fit into the modular GRADD architecture.
The bigger problem I see is the fact that the OS/2 Presentation Manager and all the graphical OS/2 applications use the GPI API to draw something on the screen. If an application asks for a circle to be drawn, it has to be boiled down to the graphics card somehow, right?
Now, when IBM decided to throw out the old video driver architecture, and introduced the much more clear GRADD architecture instead, they have written the "transition layer" which connects the GPI API with the GRADD drivers (gre2vman.dll, anyone?), so the old OS/2 PM applications did not have to be changed, they ran without modifications on the new video driver architecture.
I'd just like to point out that, in case the whole GRADD architecture is going to be thrown out, somebody will have to write a new transition layer, which will be quite a huge work, including a lot of gray and undocumented areas of the OS/2 PM API. Combining this with my non-understanding why GRADD does not fit for GL/2, I just don't see why was this path chosen.
Again, I'm absolutely not against GL/2's current plans and directional decisions (anyway, who I am to be?), these are just my thoughts on the case.
And please let me have one more sentence: as I understood Rudi's posts, I think he was also quite interested in GL/2, and tried to understand some of the decisions (having similar problems like me) to be able to help... too bad that some misunderstanding or miscommunication or whatever came into the picture...
Doodle