I've advised "flashman" by email about the problem but I thought everyone might be interested in the cause.

1.02-rc1 (and a few betas earlier) does some better disk full detection. It has a safety margin that is required for rebuilding the database and other stuff. If a player has already exceeded the safety margin when 1.02-rc1 is installed then it will produce disk full errors for most synchronisation tasks - in 1.1 this should be dealt with much more cleanly.

1.02-rc1 had a bug in the disk space detection code whereby it would wipe the contents of the file before checking how much space was available - so, if you renamed your root playlist so that the new name was longer than the old name (or you introduced some new tags that weren't there before) then it would erase your 101 file, decide that there wasn't enough disk space and then give up. When rebuilding the database the root playlist would be ignored since the 101 file didn't exist and everything would end up in Unattached Items.

Needless to say that I've made it check the disk space before it wipes the file and the disk full error now actually says "out of disk space" so that people will at least have a clue to the problem.

I've trashed my root playlist several times today during testing and so far it has just been a simple task of dragging everything back out to the root again. This morning we worked out a better unattached items collection algorithm that should make this even more reliable.

--
Mike Crowe
I may not be speaking on behalf of empeg above :-)
_________________________
--
Mike Crowe