30 messages in com.perforce.perforce-user[p4] p4 client stdout buffering chang...
FromSent OnAttachments
Rick Macdonald06 Apr 2000 23:20 
Jonathan Biggar06 Apr 2000 23:37 
Steve Cogorno07 Apr 2000 08:16 
Richard Geiger07 Apr 2000 08:45 
Steve Cogorno07 Apr 2000 08:53 
PETERS MW Mr -NUCLEAR07 Apr 2000 09:02 
Richard Geiger07 Apr 2000 09:27 
Richard Geiger07 Apr 2000 09:28 
Jonathan Biggar07 Apr 2000 09:34 
Steve Cogorno07 Apr 2000 09:39 
Richard Geiger07 Apr 2000 09:50 
Steve Cogorno07 Apr 2000 09:56 
PETERS MW Mr -NUCLEAR07 Apr 2000 10:03 
Richard Geiger07 Apr 2000 10:18 
Richard Geiger07 Apr 2000 10:28 
Scott Blachowicz07 Apr 2000 11:42 
Rick Macdonald07 Apr 2000 13:45 
Scott Blachowicz07 Apr 2000 13:59 
Jonathan Biggar07 Apr 2000 14:01 
Schmidgall, Jay07 Apr 2000 14:05 
Richard Geiger07 Apr 2000 14:13 
Rick Macdonald07 Apr 2000 14:19 
Rick Macdonald07 Apr 2000 14:43 
Scott Blachowicz07 Apr 2000 17:55 
Rick Macdonald07 Apr 2000 18:13 
Scott Blachowicz07 Apr 2000 18:28 
Rick Macdonald07 Apr 2000 18:38 
Christopher Seiwald10 Apr 2000 14:56 
Rick Macdonald10 Apr 2000 15:07 
Rick Macdonald11 Apr 2000 11:35 
Subject:[p4] p4 client stdout buffering change request
From:Richard Geiger (rm@netapp.com)
Date:04/07/2000 10:18:55 AM
List:com.perforce.perforce-user

Richard Geiger said:

I actually suspect the problem is really that all of the buffer flushing is done in the stdio libary anyway, and no write() system call to flush the buffer is being made until it's full. Can't do much about that down in the kernel, can we?

Oh. Does perl write using stdio? If so, then yes, there isn't much the kernal can do. Forgive my perl ignorance - I've never used it.

Proper buffering applies to direct syscalls like write/read etc.

I'm guessing that Rick's original request is based on an assumption, quite possibly correct, that "p4" uses stdio; It would thus seem to be a very reasonable request. You may have misconstrued with the "please don't do this" reply, I guess.

The suggestion that Steve connect via a pty to get unbuffered behavior was a good one, too, allowing him not to require any changes on the part of Perforce.