>as far as i was able to determine, there were two distinct points where
>serial performance got worse.  once between oct 3 and oct 19, 1996, when
>i started seeing a few more overruns, and then once around dec 31, when
>my serial port became unusable at 38400 (IPX).

I know this has been brough up before ... but can't someone with CVS
access _list_ all of the changes between these dates and thus narrow down
the possible causes of the problem?

--Ken