From: Stuart Tedford <stuart.tedford_at_piresearch.co.uk>
Date: Wed, 5 Feb 2003 10:04:35 -0000

This was discussed a few months ago. The main outcome was that the
next_buffer could be overwritten with the current_track after the first few
seconds of current_track play, with no skip button pressed, and while the
HDD is still spinning.

The only real problem with what you proposed is that a big chunk of the
current buffer is wasted for the first spin-up/load/play cycle, by keeping
the next_buffer alive long after those first few seconds. Thus causing the
next load cycle to occur sooner.

The solution would be a bit of a compromise becasue you don't want to keep
the HDD spinning any longer than normal, but you want to give the user a
chance to decide if they want to skip. Maybe the skip_detect_time should be
the same as the existing HDD spin down time.

It also won't be a "five second job(tm)" (my bosses favourite saying). It
might be better to spend that time making the skip button detection better
when loading the current track, to reduce the latency the user experiences,
which I agree, is annoying. But is is a damn sight better now than it was
back then, so I don't really get annoyed by it anymore, and certainly not
annoyed enough to show anyone any code ;-)

Stu.
Received on 2003-02-05