Geoff Winkless wrote: > Why not a tagged format and tools which can handle both types?
If a new image format is going to be created out of all of this, it would certainly make sense to be flexible enough for both uses. A tagged format would be easy to process and nicely extensible too. Beyond a file signature, should anything else be mandatory? Having a fixed disk geometry in the header has always seemed a bit wrong to me, as real disks don't impose such limits (beyond the physical movement of the head, and the track bitstream length). Only formatted areas need be stored too, ideally with a special case for efficient storage of simple/regular formats. Descriptions and CRCs could be optional tags, and there's probably not much else that really belongs in a pure disk image (though they are easily added). A new unformatted disk would just need the few bytes for the signature itself, and perhaps an end tag, since there's no other data. A freshly formatted SAM disk would require just one tag (plus a few parameters in data) to tell it to start with a pre-formatted image with 2 sides, 80 tracks, 10 sectors/track, filler byte 0x00. A near empty disks would use the same, but would be followed by any data that is present (haven't thought about details yet). Fancier disks would contain track entries for any formatted ones, each containing a sector count and list of sector headers. The program creating/saving the image would be responsible for picking the most efficient/suitable method for storing the image, and in most cases will be the pre-formatted+data case from above. Programs reading the images would need to determine whether the stored format is suitable for their needs too, such as converting to other limited formats like DSK. This all assumes that the idea of a new archive image format is for "PC"-side storage and transfer of disks, and that images will typically be created and written on the PC-side too. There are no special considerations storing/using the images on the SAM itself, as that feels like a separate application-specific issue. Good, bad or ugly? Si
