I'm not sure if this is actually a bug but the output settings in the Device tab have no effect on the resulting WAV file created by WAV Writer.
For example, if you set the device to WAV Writer, set it to Mono/22050/8 (for whatever reason) then play a Stereo/44100/16 file, the WAV file will also be Stereo/44100/16 instead of Mono/22050/8.
Tested here with 3.2.0.6 (28 april, 254556 bytes). Works as expected with a MO3. With a MP3, only the bit depth setting is preserved (rest is whatever the file was encoded at), which makes sense in a way (the Device tab has a note saying that the sample rate only affects MODs).
If you start playing a MIDI file then stop it, it will start playing from start of the file again but this time the XMP display (time counter, VU meter, etc) don't run. (Yes I made sure that it's a MIDI not a MOD, even so, I still made sure there was NO loops active).
Are your MIDI settings configured correctly? I'm assuming you're using the winamp plugin. It needs to be set to use DirectMusic and to use winamp's output, otherwise you'll have problems.
Testing here (with same version)... I've noticed that when you stop the MIDI file the plugin doesn't reset the MIDI channels. So if you then click play without unloading the file first, it'll finish playing the bar you stopped it at and then start from the beginning. There's also a similar effect when seeking, probably due to how the plugins handles seeking in MIDIs.
Possibly related to the above bug, XMPlay uses very little CPU (~0-1%) normally, but as soon as you start playing a MIDI file, the CPU usage jumps to about 20-25% and stays like that until you quit XMPlay.
Here XMPlay uses 5-10% (Athlon 1.3GHz) when a MIDI file is loaded and/or playing, and that drops to about 0% when the file is unloaded (press Stop twice to unload a file). Again, check your plugin settings.
Of course this could all have appeared in one of the new builds of XMPlay. Ian, could you start adding build numbers to XMPlay?
Edit: Typoed. It's "bit depth" not "bitrate" that's preserved.