The following is a conversation I had with the
serial.c maintainer on the subject of serial
overruns. I believe *any and all* errors should be
returned via read(2), but apparently ignoring
overruns is the accepted practice.
(Just thought this might be helpful to someone
else.)
---------- Forwarded message ----------
Date: Mon, 1 Jun 1998 20:18:19 -0400
From: [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Subject: Re: Problem catching TTY_OVERRUN
> Date: Mon, 1 Jun 1998 08:49:05 -0700 (PDT)
> From: John H Darrah <[EMAIL PROTECTED]>
> I have encountered a problem where read(2) does
> not return an error on an overrun condition. Is
> this normal or is there a termios function that
> enables this?
Yes, this is normal. Normally, overruns are treated merely as dropped
characters, and most modern protocols already have ways of noting the
problem (checksums, etc.).
Usually overruns indicate a system misconfiguration, and should be fixed
by the system administrator. That's why I've never bothered to have a
way for the application program note overruns. This is pretty standard
across all Unix and Unix-like systems.
- Ted
-------- End Forwarded message ---------
--
John Darrah (u05192) | Dept: N/C Programming
Giddens Industries |
PO box 3190 | Ph: (206) 767-4212 #229
Everett WA 98203 | Fx: (206) 764-9639
--
PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST ARCHIVES!
http://www.redhat.com/RedHat-FAQ /RedHat-Errata /RedHat-Tips /mailing-lists
To unsubscribe: mail [EMAIL PROTECTED] with
"unsubscribe" as the Subject.
