Playing with Kryoflux

[Keir]

Hello all! I took delivery of a Kryoflux yesterday and have been testing it out. Down below there's a picture of it munching on a Lemmings original game disk.

I'm happy so far, as it saves getting out the old Amiga 1200 to rip disk images, and I'm finding significantly fewer errors when I post-analyse the dumps.

My new workflow is to dump Kryoflux STREAM files with half-dozen revolutions per track:

dtc -r6 -flemmings_disk1_ -i0 -e79 -i5

Then I convert these to my own dumb format containing raw bit cells plus timing information (1 byte of MFM plus 1 byte of Amiga CIA ticks taken to read that byte):

kf2rawdat lemmings_disk1_ lemmings_disk1.dat

And I have tools to then convert to ADF, to extract Amiga filesystem into a host directory, or to pick out other custom formats (the interesting case ):

mfmparse2adf workbench.dat workbench.adf
adfread workbench.adf

I've attached some tools for those who are interested. At least the parsing of raw STREAM files may be interesting, as there's a modicum of trial and error to apply the rules in the published specification.

EDIT: Removed the archive from here. The up-to-date version of the archive can be found over at the Kryoflux forums. See fiath's link below.

[mr.vince]

Awesome! Good stuff!

[prowler]

Originally Posted by kaffer

I've attached some tools for those who are interested. At least the parsing of raw STREAM files may be interesting, as there's a modicum of trial and error to apply the rules in the published specification.

Hi kaffer,
Nice work! Any chance of some binaries, please?

[fiath]

Wow, excellent work. Good to know it is possible to make an independent implementation of the spec too. If you have any suggestions on improving the spec, they would be very welcome. Maybe there are sections that need expanding on?

[Keir]

My main confusion was over stream positions, and where that could mean byte position in the stream file, offset in the decoded stream of flux reversals, or even offset in the implied stream of bit cells (the spec mentions cell positions which I found confusing, as I've only seen the term used in the context of bit cells, which doesn't make sense when decoding raw flux transitions without assuming further processing/analysis).

This meant I took a guess at what the 'stream position' field in OOB messages means. No matter how I interpreted them, I couldn't get the position field in successive 'stream read' messages to line up exactly. I ended up assuming that the positions are offsets in the stream of flux reversals, as that matched closest and most consistently, but still I *always* find that the stream-read message indicates an offset three or four transitions further ahead than I've decoded. I'm guessing the stream is in fact continuous per-track, with no interruptions, so I've got something wrong somewhere, no doubt!

[fiath]

For anyone interested, the response to these queries has been posted over in the KF forum, which is probably the best place for any further technical discussion.

http://forum.kryoflux.com/viewtopic.php?f=7&t=199

[r.cade]

disregard.