Decoding Photographs of the IBM 5100's Executable ROS

github.com

36 points·mmastrac··8 comments

> The youngest 5100s are a bit over 40 at time of writing, and some accounts online suggest that the ROS devices are no longer dependable.

That must be why John Titor needed to go back to 2000 to retrieve one.

Well, I'm glad this was the first post. Let the legend live on.

I guess going back to 2019 and getting a copy of the ROS dump from Github was too easy ;)

Havent heard that name in a while!

Why not just grab the data with a logic analyzer while it's on the bus? I see no reason for the whole OCR workflow here.

I'm guessing the OP didn't have a logic analyser.

You'd need to grab the address and data buses, and synchronise that to a point in the fetch cycle when the data is guaranteed to be stable.

It also means a lot more test clips to connect... A 16-bit bus with 16-bit addressing would be 33 data clips and a couple of grounds. In a cardcage type system you'd need an extender card to do this. Without an extender card you need to solder in test pins or wires. Not something I'd want to do to a 40-year-old museum piece.

There's a picture in the repo it looks like he is using a logic analyzer. He also mentions the backplane is wire wrapped, making it easy to connect wires.

https://github.com/stepleton/5100ExecutableROSDecode/blob/ma...

They probably could have also used an arduino with some circuitry to automate the stepping and address line management, as well as read the data lines. This assumes of course that the PALM has such systems.

My guess? The writer had OCR on the mind and decided "Why not?"

Reply on news.ycombinator.com