[sldev] DirectX - feasible to replace?

[Boroondas Gupte]

[oops, used wrong address ... here it is again for the list]

Ambrosia schrieb:
> Given that that one doesn't use the DirectX SDK for
> hardware detection, have there been certain cases where the Linux
> viewer failed to detect hardware properly, with negative results other
> than a simple 'can't detect X' message?
Hmm ... what exactly should we watch for? SL's automatic VRAM detection
was off by a factor of 2 on linux, somewhen, but I think that's been
fixed by now. SpaceNavigator works fine with builds from LL's current
public SVN trunk, which includes a linux drop-in replacement for the
(windows-only) libndofdev. The only issues there are that the user
account running SL needs access (rw-permission) to the device, which is
best provided by udev via some group (e.g. usb or plugdev), and that the
drop-in (which is required by SL to use the SpaceNavigator) cannot be
used in parallel to the official driver from 3Dconnexion (which is
required by some other applications to use the SpaceNavigator). As the
drop-in just replicates the original library's API, but does work quite
different internally/at the back-end, I guess there will be very
different issues on Windows with SpaceNavigator and Joysticks when
removing/replacing DirectX.

What doesn't work well (or for some, not at all) under linux, is the
detection of available sound IO devices and sound
servers/systems/daemons. I don't know if DirectX is involved with this
topic in SL under Windows, though. Sound handling under linux is overly
complicated due to historic reasons and the inhomogeneity of different
distributions in this concern. Most applications handle the situation
better than SL does, but the majority of them have the advantage of
being provided by the distributions and thereby coming in versions
compiled specifically for the sound system(s) used by a specific
distribution.

What is annoying, is that sound detection seems to work different for
vivox and the rest of the viewer (whether fmod or OpenAl or both). While
it makes sense to have separate settings, so you can e.g. output normal
sound on the speakers and voice on the head-set's headphones, it is very
confusing to have different sets of detected devices available for the
two settings, or seeing different names for the same devices.

cheers
Boroondas