I'm sure vxwindows is just fine. As for the why, well this project is a pet of myself and some like minded friends who only ever wanted to write games :-)
Actually, what were is aiming for is to create a low overhead C++ interface to common OS features. It isn't aimed at providing high level interfaces (ie encapsulating the OS GU), rather it is designed for performance. In other words, gaming and multimedia :-) Similar in intent to SDL but aimed soley at C++.
There are several libraries, in different stages of completion:
systemlib (memory, threading, locking, timers etc.)
gfxlib (pixels, surfaces, displays, blits, 3D rasterizer, drawing classes etc.)
iolib (file io, input devices, event filtering etc.)
guilib (abstract gui control framework, experimental)
soundlib (not started yet)
netlib (not started yet)
The code is split into 2 parts - system independent code and system dependent code. Only truly OS independent code is allowed in the former section for cleanliness. One of the design aims is that no system defined behaviour or data is present in the user APIs. Code using only the provided classes (and derived from them) should ideally compile 'out of the box'.
The latter section of the code has an OS tree and the implementation level classes are free to utilize as much of a given OS's functionality as they can to provide performance as long as the overall expected behaviour is maintained.
Naturally, the amigaos tree is the most developed, but there is a guy working on the winblows2k part of the OS tree and another who hopefully will start the macosx part in the not too distant :-)
This brings me to my current problem. Currently all the work has been done in storm but it is being moved to gcc 2.95.3 as a minimum compiler spec largely because storm lacks some features (reliable templates, namespaces, thread safe exceptions etc) that caused me to make some awkward design descisions a while back.
By moving to gcc I aim to clean up some of the encapsulation (using namespaces) and also to make the code more accessible on other platforms.
Once the whole project is at an acceptable state of completion no doubt it will go open source ;-)