Hacker News Reader

Comment by h2odragon | Hacker News Reader

Right. We need to go back to parallel ports. Simple, 25 pins on one ends, 25 or 36 pins on the other, only a few handshake lines and data lines that were data lines not some polymorphic "who knows what you're gonna get" portal to RICH CHUNKY AMPS.

Of course there was the same variety of standards and connectors and in some instances even incompatible electrical implementations there too; just without the advertising of "everything is supposed to work together".

maybe our progress has been limited to the propaganda side alone.

View on HN

I definitely don't miss the infinite combinations of DB-9 to DB-25, making sure your TX/RX lines were lined up correctly, baud rate, 8N1 (or whatever), etc. Then making sure your UART can actually do it all...

But is it really fair to compare 80s/90s connectivity to 2020 while skipping over the decades between 2000 and 2020?

cough that’s DE-9 you’re talking about; the B shell is large and the E shell much smaller.

"Baud rates" woulda been rs232 serial, unless you were getting really advanced in your parallel port fiddling. rs232 is still recognizably in use in surprising places, often on bare header pins, but still there.

"fair?" I dunno. the problems recur, the intermediate period had "VGA standards" that flowed over a wide range; and others that exhibited the same problems we see with USB-C here.

I'm pretty sure they had similar issues with crap like antenna connectors and tube socket standards back when they were the cutting edge, too.

The field I'm in still uses serial a ton. It does what's it's supposed to and it's plenty fast how it is, so no has changed it in decades.

I don't remember all of that RS232 stuff because my memory is that good - I'm reminded every time I deal with anything "embedded" :). I still cross my fingers when flashing - hoping my chosen speeds and params won't brick the device.

I helped design an USB-to-serial converter thing to directly go into the silicon of one of these chips you can program. It acts like an USB-to-serial converter on the USB side, but on the device side, it 'forgets' the to-serial and simply directly spits out the bytes it receives on USB. It's interesting how many serial-related parameters become no-ops - you can set the baud rate, parity, data bits etc to whatever value you want, the device happily ignores it and everything keeps working anyway. It's the nice part of tight standards, I guess.

(But don't ask me about the shenanigans Windows needs to see in the USB descriptor to allow programs to access an otherwise driver-less USB device - yech)

I definitely remember the difference but other than the various high speed modes (EPP and EPC, from what I recall) parallel ports worked for their intended purpose.

I mentioned RS232 serial because that was a true "cluster f**" with the issues I enumerated and more.

VGA was fun too. I remember early Linux documentation materials warning users that the wrong scanline modes for X could fry the hardware. Not sure if that was an actual occurrence but I remember triple checking just to make sure :).

Reply on news.ycombinator.com