Date: Thu, 17 Jun 2004 13:03:13 -0400 From: "Dan Langille" <dan@langille.org> To: Roman Neuhauser <neuhauser@chello.cz> Cc: ports@FreeBSD.org Subject: Re: HEADS UP - master/slave ports Message-ID: <40D19691.26918.7BC6780D@localhost> In-Reply-To: <20040617130457.GD2664@isis.wad.cz> References: <40D159DD.23350.7AD95699@localhost>
next in thread | previous in thread | raw e-mail | index | archive | help
On 17 Jun 2004 at 15:04, Roman Neuhauser wrote:
> > > and I'm not sure that's any better than what you had originally.
> >
> > Why do you think that?
>
> Because I don't know whether two /bin/realpath calls are cheaper
> than one /usr/bin/sed.
>
> > > But, what will both versions output in databases/mysql40-client?
> >
> > Both output the same thing:
>
> Ah, sure, ${MASTERDIR} was passed through realpath in both versions.
>
> So, remaining differences are perhaps speed, readability, and more
> than one level of categories. The first hasn't been measured, the
> second is a subjective thing, and the last is ATM a purely
> theoretical concern.
FWIW, I created a script which invoked the command 15,000 times.
The original patch:
real 15m54.561s
user 10m40.866s
sys 5m1.405s
The patch derived in this thread:
real 15m48.481s
user 10m39.745s
sys 4m54.753s
Confession: Yes, this isn't a very subjective test, but it's a place
to start from if someone is concerned about the performance issues
herein.
--
Dan Langille : http://www.langille.org/
BSDCan - http://www.bsdcan.org/
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?40D19691.26918.7BC6780D>