Date: Fri, 14 Aug 2009 12:57:42 -0400 From: John Baldwin <jhb@freebsd.org> To: Tim Kientzle <kientzle@freebsd.org> Cc: "Bjoern A. Zeeb" <bz@freebsd.org>, freebsd-current@freebsd.org, Doug Barton <dougb@freebsd.org> Subject: Re: svn commit: r190514 - head/sys/conf Message-ID: <200908141257.42672.jhb@freebsd.org> In-Reply-To: <4A8583DB.1090507@freebsd.org> References: <200903282317.n2SNHIjI015202@svn.freebsd.org> <200908141004.09354.jhb@freebsd.org> <4A8583DB.1090507@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Friday 14 August 2009 11:33:47 am Tim Kientzle wrote: > John Baldwin wrote: > > On Thursday 13 August 2009 2:57:10 pm Doug Barton wrote: > >> Bjoern A. Zeeb wrote: > >>> Author: bz > >>> Date: Sat Mar 28 23:17:18 2009 > >>> New Revision: 190514 > >>> URL: http://svn.freebsd.org/changeset/base/190514 > >>> > >>> Log: > >>> For kernel builds reduce the impact of svnversion, just scanning > >>> src/sys and not the entire src/ tree. > > Performance here I think is a red herring. This is > really about correctness: The SVN revision of usr.bin/ls > simply isn't relevant for the kernel build. Very true. > >> Also, what problem are we really trying to solve here? With a > >> populated cache it takes on average 5 seconds to run all of src, and > >> just under 1 to do only sys. Is 4 seconds really that important to > >> save? With a dry cache I'm sure it takes a little longer, but has > >> anyone actually measured this? > > I just measured over 30 seconds for svnversion against /usr/src and > around 6 for /usr/src/sys (both with cold cache). > > > It takes far longer than 5 seconds here against a local SVN repo over NFS. > > The repo has nothing to do with it. svnversion doesn't > talk to the repo. It only examines the working copy. Ah, true. My checkouts are also over NFS though rather than local disk which may explain it still. -- John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200908141257.42672.jhb>