Hacker News Reader

Comment by jwr | Hacker News Reader

I have a keyboard from my first 386/33 computer. It can be connected to my current Mac through a sequence of adapters: DIN5 -> PS/2 -> USB-A -> USB-C.

I am looking forward to extending my sequence of adapters in the future.

View on HN

Windows 11 will happily load sermouse.sys if you hook a serial mouse to a USB serial adapter and plug it in.

Even on arm64!

Now that’s an old-new-thing post I’d like to read!

I assume because serial mouse support ca. 1996 is baked deeply into the USB standard, so you can’t pass connect-a-thon/compliance tests without it?

It might also just be a situation of "it's there, doesn't hurt anyone, doesn't cause much work, removing it could hurt someone". Windows needs (or at least used to need) keyboard support earlier in the boot sequence than mouse support. It's plausible that they load the entire USB stack early on to get keyboards etc working, then sometime later some ancient code iterates all serial ports to check for serial mice. To that driver, USB-Serial bridges and "native" serial ports should look the same through the hardware abstraction layer.

The leaked Windows XP code contains a lot of ancient code, some apparently taken from MSDOS. They later removed some of that, but apparently not this part.

> It's plausible that they load the entire USB stack early on to get keyboards etc working, then sometime later some ancient code iterates all serial ports to check for serial mice. To that driver, USB-Serial bridges and "native" serial ports should look the same through the hardware abstraction layer.

That's exactly it. When a serial port is initialized on Windows, it looks for data coming in, and if the right 'garbage' at the right time is picked up, Windows will then hook the mouse driver.

https://roborooter.com/post/serial-mice/ discusses the fairly simple protocol at hand.

This can lead into the unintentional effect of "plug in <serial-based device, like a GPS>, mouse suddenly goes wild". In my case... works as designed - serial mouse on USB-C dongle just works ;)

https://www.taltech.com/support/windows_2000_nt_serial_mice_...

It isn't. Serial mouse protocols are nothing like USB HID protocol.

I have something similar at my desk, an Apple Extended Keyboard II. It is connected to an ADB to USB adapter (which needs its coin cell battery replaced every 5-7 years) and then that is connected to a USB-A to USB-C adapter to connect to my MacBook. Despite being about 35 years old, it still functions beautifully and I use it every day. I really hope that I don’t have to add another adapter to my chain in the future but you never know…

I would think the PS/2 -> USB-A adapter in this chain is a smart one.

Um. Yes. It must be, or it would not work.

Yes, of course.

I still had my DEC LK201 hooked up on my main home linux desktop until a few years ago. Haha. I don't know why, but I liked the key layout.

Can't you get a DIN5 -> USB-A adaptor? Not sure if that would work, might be something designed for audio devices.

AT keyboard (DIN5) and PS/2 keyboards are in terms of electrical signals and the low-level framing protocol equivalent. As for the higher protocol level PS/2 is technically an superset of AT keyboard, but when used on PC platforms (or with Linux) the difference does not matter. And in reality most of the cheap keyboards you can buy only implement small subset of the protocol that is used by BIOS and Windows, such keyboards will not work with things like DEC Alpha workstations. (this also leads to somewhat surprising observation that under windows some keyboards work when plugged into mouse port, some do not)

But where's the fun in that?

> I am looking forward to extending my sequence of adapters in the future.

Keep us updated about how long it gets. :D

Pics or it didn't happen. (Seriously, I'd love to see that unholy train.)

Reply on news.ycombinator.com