Mailing-List: contact cygwin-help@cygwin.com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe@cygwin.com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin@cygwin.com>
List-Help: <mailto:cygwin-help@cygwin.com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-owner@cygwin.com
Mail-Followup-To: cygwin@cygwin.com
Delivered-To: mailing list cygwin@cygwin.com
Date: Wed, 11 Dec 2002 14:35:53 -0500
From: Christopher Faylor <cgf@redhat.com>
To: cygwin@cygwin.com
Subject: Re: pipe improvements in snapshot
Message-ID: <20021211193553.GA30532@redhat.com>
Reply-To: cygwin@cygwin.com
Mail-Followup-To: cygwin@cygwin.com
References: <20021211041250.GA31215@redhat.com> <44125937.20021211150845@huno.net> <20021211192639.GE29798@redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20021211192639.GE29798@redhat.com>
User-Agent: Mutt/1.5.1i

On Wed, Dec 11, 2002 at 02:26:39PM -0500, Christopher Faylor wrote:
>On Wed, Dec 11, 2002 at 03:08:45PM +0100, thomas wrote:
>>Christopher Faylor <cgf-cygwin@cygwin.com> wrote:
>>
>>> Please check out the latest snapshot and report here if there are
>>> problems.  I haven't yet tried this on Windows 9x class systems so it's
>>> entirely possible that there is a problem there.
>>
>>It seems to work great! I did a few tests and there was no delay anymore
>>whatsoever. I've just sent the dll to someone to try out on a 9x system.
>
>I'll test this on 9x myself eventually.  With gritted teeth...
>
>>One thing about the possible data loss: Is that true data loss, like
>>some bytes won't make it trough the pipe, or will that only result in a
>>delay because the bytes have to be send again?
>
>It's true data loss but it is very very unlikely that it will ever be hit.
>I thought I would get this out there while I ruminated on ways to eliminate
>the potential for loss.
>
>>I will do some more thourough tests and will report back.
>>
>>Thanks so far!
>
>You're welcome.

Btw, thanks for the feedback.  It's an interesting datapoint.  It seems
to mean that the context switch implicit in introducing a thread into
the pipe read is substantially less than the 10ms delay we were adding
in the old scheme.  This probably makes sense but...  this is Windows,
so....

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

