[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Nmh-workers] m_getfld
From: |
markus schnalke |
Subject: |
Re: [Nmh-workers] m_getfld |
Date: |
Mon, 10 Dec 2012 08:55:43 +0100 |
User-agent: |
mmh 0.0-dev |
[2012-12-10 08:40] markus schnalke <address@hidden>
> [2012-12-09 18:59] Ken Hornstein <address@hidden>
> > >Yes, but by "different purposes" I was thinking how scan
> > >digs back into the IO buffers. Though maybe that won't
> > >be necessary any more.
> >
> > I had forgotten about that. Okay, looking at that now .... alright,
> > that's not as nasty as I thought. All it does is to use the output
> > stdio buffer as the input buffer for m_getfld() so we can avoid an extra
> > copy. Can we all agree that's not necessary anymore, and the resulting
> > performance gain is probably miniscule? If so, I can simply get rid of
> > that garbage now.
>
> Just one word about performance:
>
> I've removed the ``dig into internal IO buffers'' stuff in March and
> noticed no performance loss. See
> http://git.marmaro.de/?p=mmh;a=commitdiff;h=d7b4f0034bc4f5b5c2f990d0984858e9b6f4131a
Ups sorry. That was just one of the several ``dig into interal IO
buffers'' places. Sorry for the noise. :-/
meillo
- Re: [Nmh-workers] m_getfld, (continued)
Re: [Nmh-workers] m_getfld, David Levine, 2012/12/09
Re: [Nmh-workers] m_getfld, David Levine, 2012/12/09
Re: [Nmh-workers] m_getfld, David Levine, 2012/12/09
Re: [Nmh-workers] m_getfld, David Levine, 2012/12/09
Re: [Nmh-workers] m_getfld, David Levine, 2012/12/10