joegorin wrote: 
> As I understand the proposed experiment, I could reduce the library size
> to a small size, such as a few album files, and run the machine for more
> than five days. If it runs for a week, then I might change the library
> to half of its current size and run again, then 3/4 or 1/4 size,
> depending on whether it fails, etc. This is a very long experiment but I
> think I will start it.
> 
> The "bad file" hypothesis is consistent with my observation that a
> simple search does not find a lot of people having this problem. But the
> opinions of experts on this thread makes me not optimistic.

After thinking about it, I reject the bad file hypothesis without
further experiments. My logic:

Assume there is exactly one bad file out of 40 hours of music. The
probability of coming upon that file when randomly shuffling is 63% in
40 hours, and 95% in five days. In other words, the chance that I will
not come upon it before 5 days is only 5%. The probability that I
observed failures five times somewhere around 5 days apart is 3 chances
in 10 million. For a hypothesis of more than one bad file, the chance of
my observations is even smaller.

The "memory leak" model makes more sense.


------------------------------------------------------------------------
joegorin's Profile: http://forums.slimdevices.com/member.php?userid=68547
View this thread: http://forums.slimdevices.com/showthread.php?t=110255

_______________________________________________
Touch mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/touch

Reply via email to