From: Johnny Billquist <bqt@softjar.se>

>And then came IBM, with it's typical fashion of just breaking 
>everything, and moved to a DE-9 with the wires totally messed up, so 
>that all bets were off again. *sigh*

It could be worse, at least the DE-9 was male!  And that was (originally)
only COM2 on the AT dual-serial board (since the back plate had space for
DB25+DE9 but not for two DB25s).  The AT's COM1, and the PC/XT COM board,
used a male DB25 wired as DTE (although the PC/XT original also used some
pins for 20mA and had a jumper plug to change modes), so things started
out OK at least, not like the Radio Shack junk which was all backwards.

I really don't get how people decided the DE-9 should become the first
choice though (I think it was serial mice that started it, they were usually
on COM2).  I still have piles of DE-9/DB25 cables, and even modems made
for PCs (Hayes, USR, Bocamodem) usually had DB25s on them (until later).
It's surprising that IBM's DE-9 pinout didn't happen to look more like
DEC's, I mean it seems like the obvious thing to do to change as few pin
numbers as possible compared to the DB-25 so you'd think a data-leads-only
2+3+7 cable would have worked everywhere.  It sounds like IBM was really
cheaping out on development though, they honestly thought PCs were a fad
that wouldn't last (that's why they chose the 8088 in the first place --
it worked with 8080A peripheral chips so they could slap something together
w/o having to cook up an I/O system from scratch).

John Wilson
D Bit