2022-11-20 13:08:05 <KipIngram   > Ok, so I've refined the part of the pseudocode that will be the script that runs on the master and slave canisters (one instance of it for each drive).  Once I finalize that and get it working, I'll be able to manually move work through the system, with me playing the role of the "controller."  The controller software will be a little more involved, and that's where the Python / urwid stuff will come in.  But
2022-11-20 13:08:07 <KipIngram   > I'll be able to do production work with just this easy part with a level of effort similar to what I invest now.
2022-11-20 15:54:53 <KipIngram   > dave0: Are you doing anything in particular with the Z80, or just reading up?
2022-11-20 20:54:19 <KipIngram   > So, Python advice question, if any of  you have worked with it.  For this controller, I want to be able to do very good recovery from failure events (think power failure).  So, at appropriate times I will have the controller serialize all of its data, which will reside in a single integrated data structure, to disk.  I'll include a hash for integrity checking.  Then I'll be able to "restart" and it will read
2022-11-20 20:54:21 <KipIngram   > the structure back in and pick up where it left off.  The master and slave will keep all of their important data in files referenced by bash.
2022-11-20 20:54:36 <KipIngram   > So, any reason to prefer json vs. pickle for that serialization / deserialization?
2022-11-20 20:54:59 <KipIngram   > I lean toward json because it's human readable, but if pickle has some other advantage it might outweigh that for me.
2022-11-20 20:55:25 <KipIngram   > I have more experience with json too, and it's just so damn easy to use.
2022-11-20 20:56:40 <KipIngram   > I'm going to keep the last several serializations around, so that if the latest fails hash check I can just drop back to the one before that.  At worst it will just cause me to re-run a test step I otherwise wouldn't have to.
2022-11-20 21:50:57 <KipIngram   > Oh my - there is an extension of pickle, package dill (people are so cute).
2022-11-20 21:51:13 <KipIngram   > It is explicitly designed to be able to save and restore entire interpreter sessions.
2022-11-20 21:51:25 <KipIngram   > dill.dump_session(<filename>)
2022-11-20 21:51:29 <KipIngram   > and you're done.
2022-11-20 21:52:15 <KipIngram   > So all I have to do is write this so that the initial value of a state variable determines where in the logic the thing kicks off, rather than just "proceeding from an entry point."
2022-11-20 21:52:58 <KipIngram   > It'll always enter the "big loop" from the top, but variables will control what it actually does.
2022-11-20 21:53:46 <KipIngram   > Then once that's working, the next step will be to arrange for it all to start on system power-up automatically.
2022-11-20 23:32:13 <KipIngram   > Hey, did you guys know that if you 'import code' in Python, t hen anywhere in your program you can run code.interact() and it will pitch you into an interactive Python console?  Ctrl-D takes you back to your proram.
2022-11-20 23:33:12 <KipIngram   > if you pass local=locals() as the parameter string, I think  you can then query your local variables around the call, though I haven't actually tried that yet.