The maintainer of the Arm Linux client responded to the bug, he's going to release a pthreads version, thus making it compatible with the empeg again.
I figure I'll set up the client similar to the way I did back on my Mark 1. The buffers will be held in a ramdisk, as the client enjoys frequently writing to them. Then, on a timeout, the buffers will be copied from the ramdisk to the harddisk. Any suggestions on the timeout? I think last time I used seven minutes, that seemed to be a good balance to catch the short car trips, and also infrequent enough that the disks shouldn't be in rw mode too often. The actual rw time back then was about 2 - 4 seconds not including spinup time, I'm sure now that could be a bit quicker. Also, I only rwed the one partiton I had the buffers stored on, not everything.