A critical look at OpenBSD's installer
eerielinux.wordpress.com
eerielinux.wordpress.com
We seem to have gone backwards in our expectations of serial port speeds. At the height of the BBS boom, 14400 BPS was widespread, and high end modems were capable of 57600 BPS. (Yes, FOSSIL existed; and not all of us even used MS-DOS.) That 14400 BPS is no longer the expected speed that most people will do, as it was some 30 years ago, and that this is back down to 9600 BPS in 2025, is somewhat sad. We've regressed quite a lot in the computer world in several areas, especially when it comes to serial communications and terminals.
And unnecessarily so. The virtual UART in that VM was probably quite capable of Ridiculous Speed. And a cheap-o real USB to DB9 RS232 adapter based around the FT232R can do Ludicrous Speed.
Speaking of terminals: The point about "vt100" versus "vt220" is not as major as one might think. There are a few but not many features of a VT 220 that a simple block-terminal-like install program could make productive use of through termcap, let alone terminfo. (FreeBSD is one of the very few active operating systems still stuck with termcap, which is actually more significant than the default terminal type difference. And for some years FreeBSD has mis-labelled its KVTs as "xterm" and not had a "teken" terminal type in termcap, although there has been one in Dickey terminfo for quite some time, so long in fact that there has been a "teken-2022" revision.)
Not tested here, was whether one can put the install image on a DASD and then install to other partitions in the remaining free space on that same DASD. (Not everyone installs from DVD, CD-ROM, or floppy to hard disc. (-: Sometimes one has put an image onto a USB storage device or a TF card, with comparatively vast amounts of room to spare.)
From recent personal experience, I can attest that the OpenBSD installer very much does not shine in this circumstance, whereas the NetBSD installer can be persuaded to just install its "sets" to the current system. Although if one doesn't tweak the install image elsewhere before booting, it will insist upon auto-growing the one NetBSD partition to encompass the whole of the rest of the device, leaving no room for swap.
In the days of the BBS boom, our cables had all of the wires, our modems did all of the signals and had lights for them on their front panels, and our operating systems either had proper interrupt-driven access to the UARTs as standard or we used things like FOSSIL. Hardware flow control was pretty much a standard feature in those times, for consumer stuff that one could buy off the shelf. Not having it did not sell, in the times when manufacturers were pushing v.42 and then v.92 in their marketing literature.
If you are saying that now people have to worry about not having hardware flow control again, then that's further demonstration of how things have regressed. Serial cables with only 4 wires used to be something that was passé.
That said, a lot of the mythos about flow control has grown up because of conflation with the limits of access to a 1-byte FIFO 8250 UART on the ISA bus via MS/PC/DR-DOS. That was generally agreed to cap things at around 9600 BPS. But then along came 16550s with 16-byte FIFOs, and later still southbridges and hardware that just pretended that an ISA bus even existed. And OS/2 and Windows NT. Along with all of the vendors who were selling ISA or PCI internal modems.
It really never was the case that there was one fixed bit-rate that was "the fastest" above which suddenly everything changed. And nowadays the UARTs that modern PC users have (the internal ones usually secretly still being there in the chipsets) are devices on what is at minimum (and almost never) a "low" speed 1.5Mb/s version 1.0 USB. The aforementioned FT232R has 256-byte transmit and 128-byte receive FIFOs, and its datasheet claims up to 3M BPS.
Stating a serial port speed limit from roughly 1989 in 2025 is quite bizarre to those of us who lived through the BBS boom. It must have arrived in a DeLorean driven by some bloke named Marty.