Paul, at 08:26 on Feb 15 1996, you wrote to Bill Grimsley...

BG> As I said to Rod, I can't check a Rockwell, as I no longer have access to 

PE> Doesn't matter, Rod has tested them for you.

One or two, at least.

BG> any, and I don't know if they also write the port speed to NVRAM (although 
BG> I very much doubt it, otherwise they'd exhibit the same quirky behaviour as 

PE> Of course they do write it to NVRAM. 

According to Rod, non-USRs DON'T write it to NVRAM at all.

PE> It's just that unlike USR, they are not fucked by design, and do NOT use 
PE> the NVRAM speed unless NO AT COMMAND HAS BEEN ISSUED.

If it ain't stored, how the fuck CAN they use it ?

BG> ATZ restores all &W defaults from NVRAM, but if Rockwells don't store the 
BG> baud rate in NVRAM, it can't be recalled, simple as that.

PE> They do store it, it can be recalled, it doesn't get recalled, so
PE> will you now admit that the USR is fucked by design, in it's
PE> non-Hayes behaviour?

Nope, certainly not.  Especially now that I know the Rockwells DON'T store
the baud rate at all.

PE> As for 115200, if I actually try it out, I might find that com.sys in OS/2 
PE> 3.0 does support 115200, in which case I may well be going 115200.  

According to the Warp docs, NO version of OS/2 supports >57600 bps. 
You'll need to use Gwinn's SIO drivers if you want to use 115200.

I've deleted the rest, because neither of us are going to change our
opinions, and I have far better things to do with my time anyway.

Regards, Bill

--- Msgedsq/2 3.20