In article VA.0000291c.0b45b9a8@foxhill.co.uk, Ian Oliver wrote:
I reckon that over 99% of the data is present and correct, and another 0.9% will be there with good CRCs even if the headers are mangled. The remaining 0.1% will be "interesting".
OK, I just wrote 200 lines of skanky Python that does basic parsing of a DMK file, finds the id and data for every sector, runs the checksums, and writes out good data.
Despite the grim ids (some missing, those for sector 1 of head 0 for the first 34 sectors having the right checksum but wrong track number!) the data checksums were pretty much all good. However, whatever caused the ids to be wrong has clearly scambled the data. It won't restore and the mk1 eyeball suggests that one sector per track (oddly sector 15!) has duff data despite the good checksums.
All those 20 years ago, our backup software was clearly having a bad day when it wrote those disks. I've done what I can and can do no more.
Ian