--- Begin Message ---
- From: Mail Delivery Subsystem <MAILER-DAEMON fireball blast net>
- To: <gatgul voicenet com>
- Subject: Returned mail: see transcript for details
- Date: Sun, 25 Mar 2001 16:04:49 -0500 (EST)
The original message was received at Sun, 25 Mar 2001 16:03:39 -0500 (EST)
from IDENT:gat@nppp07.blast.net [207.162.129.107]
----- The following addresses had permanent fatal errors -----
<linux-sound@vger.rutgers.edu>
(reason: 550 <linux-sound@vger.rutgers.edu>... User unknown)
----- Transcript of session follows -----
... while talking to vger.rutgers.edu.:
>>> RCPT To:<linux-sound@vger.rutgers.edu>
<<< 550 <linux-sound@vger.rutgers.edu>... User unknown
550 5.1.1 <linux-sound@vger.rutgers.edu>... User unknown
Reporting-MTA: dns; fireball.blast.net
Received-From-MTA: DNS; nppp07.blast.net
Arrival-Date: Sun, 25 Mar 2001 16:03:39 -0500 (EST)
Final-Recipient: RFC822; linux-sound@vger.rutgers.edu
Action: failed
Status: 5.1.1
Remote-MTA: DNS; vger.rutgers.edu
Diagnostic-Code: SMTP; 550 <linux-sound@vger.rutgers.edu>... User unknown
Last-Attempt-Date: Sun, 25 Mar 2001 16:03:46 -0500 (EST)
--- Begin Message ---
- From: Uncle George <gatgul voicenet com>
- To: linux-sound vger rutgers edu
- Cc: scottm interlog com, sailer ife ee ethz ch
- Subject: sound under-run's screws up (struct count_info *)->bytes
- Date: Sun, 25 Mar 2001 15:43:16 -0500
I suppose the concept of handling the failure of the process to feed the
"/dev/dsp" device in a timely manner sounds ok.
The problem is that the #of bytes processed keeps going up ( and up and
up ) , even though the program has stopped feeding the i/o channel ( due
to a ^S on a printf() ) . ( SNDCTL_DSP_GETOPTR )
this makes it a little harder to keep track of the buffer stream,
because the # of bytes processed tends to now be greater than the number
of bytes written to the device. You also cant tell when the buffer is
empty ( (struct count_info *)->blocks ) because the blocks count hovers
between 0 + 1 ( i think ).
ANYWAY my expectation of (struct count_info *)->bytes is to reflect the
number of bytes processed that were derived from a user process
write("/dev/dsp" ) . The extra feeding of "/dev/dsp" due to underrun
processing should be hidden from the user.
/gat
--- End Message ---
--- End Message ---