I could probably get it done, especially if I can feed it the specific information on the Animatics controller.
Yesterday I managed to browbeat Google Gemini into creating a functional Defrag and Consolidate Free Space script for MyDefrag. It took a few rounds of telling it the error messages before it figured out why its code wasn't working. Then I posted that script to a subreddit for MyDefrag, so it's Out There for anyone looking for why this defragmenter doesn't have that core capability any HDD defragmenter must have. The why for doing that is because MyDefrag, out of the box, can either defrag all the files but not consolidate free space, or it can consolidate free space - which will fragment some files that aren't currently fragmented and may not defragment files that are fragmented. Not saying that *I* couldn't have figured that out myself, with posts on some forum - where people would fail to Read. Every. Word. of what would be my highly detailed post covering everything I had already tried, that didn't work. Might have eventually arrived at the script AI created, in a week, or more, once someone with actual reading and comprehending skills joined in. It's "The Way" of nearly every tech forum where someone will post about a problem they have, with exacting details of all their failures, and a plea for *anything different* to try - and some wag will read the title and reply with a suggestion or command to do exactly something the OP already said did not work. Then there's the corollary where the OP will never reply to anyone, there's a trail of replies that don't seem to offer anything new. Then at the end of the thread OP will reply "I got it to work! Thanks!" and vanish. That's why when I do figure out a solution I detail it, *excruciatingly detail it*. I bought a non-functional Seagate single drive Personal Cloud NAS with a bad 3TB drive. I bought a "used" drive of exactly the same model, which turned out to have around 10 hours use. Then I went hunting for the software and methods to restore it. Stuff was scattered and incomplete here and there. Some information was at least partially wrong. I painstakingly sussed out everything required and got it working and the final update applied. Then I wiped it *and did it all again*, writing a detailed procedure guide with links to download, filenames to search for if the links died, etc. And I posted that in some places where other people with a single drive Personal cloud should find it. Since Seagate shut down the Personal Cloud web services, the little box is pretty much exactly what I'd been wanting for years. A NAS with DLNA service that does absolutely nothing except "Be A DLNA Server". It cannot talk to anything outside my LAN. It needed a bare minimum of configuration, basically login with the default credentials, create a new password, disable the few non-DLNA features it has, logout. Map the thing to Z: on my PC then start pushing files to it. The only way it could be better is if it was simply a "black box" that required no login, just presented as a mappable network drive that the serves up via DLNA to the LAN any audio, video, and image files copied to it. Of course such a thing *should* have the *option* to login and do some fancy configuration, adding users and passwords, storage limits, access hours etc. But out of the box it ought to be dead simple. No arcane stuff like installing a stripped down Linux, downloading the NAS software, creating a user, installing the DLNA service, creating a shared folder, mounting the shared folder, allowing DLNA service to access the folder, locating under which completely unrelated menu one critical setting was buried. Open Media Vault 5 and 6 were all that and worse, then bogged down with a ton of useless to me bloat features that did things NOT serving video and audio files by DLNA. On Tuesday, October 13, 2026 at 02:46:02 AM MDT, Les Newell <[email protected]> wrote: On 13/10/2026 09:13, Gregg Eshelman via Emc-users wrote: > So by around May 2027 you could release your new CAM software... ;) Something like that. > > This has me thinking about the possibilities of feeding the Light Machines > PLM2000 CNC mill manual to an AI to create the Windows control software those > mills never got. I don't see why not. If the capabilities and serial command structure are documented it should be a pretty easy process. If there's a bunch of undocumented stuff going on you may have to hook up a serial sniffer. I've used AI for reverse engineering protocols before and it works pretty well. You don't need programming experience to use something like Claude Code. It helps but it's not essential. Get a claude.ai account and install Claude code. The lowest claude.ai paid tier will be plenty. Use your email as the prompt, trimming out the historical stuff and anything that isn't relevant to the direct task at hand. Tell it where to find any existing documentation files and let it run. I'd suggest not trying to integrate into LinuxCNC at this stage. One thing I have found with Claude is that it's sometimes a bit to keen to modify code. If you are just brain storming an idea and ask it a question you may need to tell it not to make any code changes. Are you familiar with Git and Github? If not, I'd suggest your first prompt just to be a very brief description of what you want to do and ask it how to set up a Github account and create a repository. It will do most of the work for you. Les _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users _______________________________________________ Emc-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/emc-users
