Trackstar disk image question

["Charlie" <charlied@NOSPAMbboard.com>]

Hi All,

I have been trying to write a routine for my FishWings program that will
allow viewing of files on a Trackstar (trackstore) disk image.

The only images I have, I downloaded probably from Asimov or Ground (I'm not
sure which).

Either these images are all corrupted (unlikely) or my routine is missing
something (almost surely true).  So here is the question:

Do any of you know the exact format that is used on the Trackstar images?

Here is what I know (or think I know):

Trackstar disk images have the ".app" MS-DOS file name extension.

They can be either ProDOS or DOS 3.3.

Everyone I have seen is 266,240 bytes long.
This is enough room for 40 tracks, each 6656 bytes long. Usually only the
standard 35 tracks are formatted.

Each track has a 128 byte header that is not encoded.  This header usually
contains some ASCII text, but is mostly filled with zeros.

The rest of the track is encoded.  The encoding used seems to be the normal
4-4 nibble encoding for the address fields and 6-2 encoding for the data
fields except that everything is written in descending order.  That is, when
reading data the last nibble of the trackstore track is the first one read.
The tracks themselves are placed in ascending order from the beginning of
the image.
The physical sectors are in a descending order that "wrap around" as if they
were on a real disk where the start of the track could be anywhere.

When read the sectors must be skewed (interleaved) of course.  I believe
Pascal skewing is used!

Trackstar images use the normal prologue and epilogue bytes but some disk
images allow one data field epilogue to be overwritten by the 128 header
mentioned above.

Inexplicably, on some images there is an occasional address field that is
written in ascending order although these seem to serve no purpose and are
not always associated with a data field.

In a normal 35 track disk there should be 560 address fields and 560 data
fields. Unfortunately, every Trackstar image that have found contains less
than 560 address fields and even fewer data fields.
Note this is using a routine that counts an address or data field only if it
has the proper 3 prologue bytes and the first two epilogue bytes.

Here is a list of the images I have (the numbers indicate the address & data
fields that my routine detects):

Name  Address Data OS

BLEEP.APP 0 0 ?
COPYII.APP 559 532 ProDOS
dosutils.app 558  530 DOS
filetrn.app 544  512 ProDOS
pdutils.app 558  530 ProDOS
SAMPLRAW.APP 555 525 ProDOS
startup1.app 0 0 ?
STARTUP2.APP 162 9 ?


Charlie