Michael J. Mahon wrote:
Since the rotational position error builds up linearly with time,
it should be sufficient to time-reference to a unique pattern on
an adjacent track, which would have a small error because of the
short seek time.
For some reason I just thought of a good reason why the image post
processing should be done (or at lest re-done) on an attached server.
You come across a disk that breaks the sync code (or something else).
You modify the code, the disk images successfully ... but does that
code still work for all the disks previously imaged? Perhaps it doesn't
matter too much, since they've already been imaged, but a regression
test against older captures would be a nice way to check. Obviously
it's going to be a bit quicker on modern computer than any Apple II.
Just a random thought.