Date: Fri, 20 Apr 2001 20:41:16 +0300 From: Ruslan Ermilov <ru@FreeBSD.org> To: current@FreeBSD.org, hackers@FreeBSD.org Subject: Re: [CFR] OpenBSD install(1) fixes: atomic install, etc. Message-ID: <20010420204116.A60253@sunbay.com> In-Reply-To: <20010420200320.A54192@sunbay.com>; from ru@FreeBSD.org on Fri, Apr 20, 2001 at 08:03:20PM %2B0300 References: <20010420191008.A51313@sunbay.com> <3AE06267.BE44852F@dante.org.uk> <3AE066C0.6B375985@FreeBSD.org> <20010420200320.A54192@sunbay.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Apr 20, 2001 at 08:03:20PM +0300, Ruslan Ermilov wrote: > On Fri, Apr 20, 2001 at 07:41:36PM +0300, Maxim Sobolev wrote: > > Konstantin Chuguev wrote: > > > > > Hi, > > > > > > Ruslan Ermilov wrote: > > > > > > > The attached patch incorporates most of OpenBSD fixes to install(1). > > > > It does not include manpage update. Most significant changes are: > > > > > > > > o New flag: -S (atomic install) > > > > > > > > : -S Safe copy. Normally, install unlinks an existing target before > > > > : installing the new file. With the -S flag a temporary file is > > > > : used and then renamed to be the target. The reason this is safer > > > > : is that if the copy or rename fails, the existing target is left > > > > : untouched. > > > > > > > > > > Just curious: why not make this way of doing install default (i.e. always use > > > it)? > > > > It may effectively doubles disk space requirements during copy (when destination > > file is not on a sofdep-enabled partition and is not open at the moment when > > install(8) unlinks it). For small files it doesn't matter, but for a big ones it > > could lead to a problem. > > > I think -S should be made the default for `installworld' (this was the main > reason that triggered this work). But I don't think this should be turned > `on' by default in install(1). > > BTW, this binary just completed the `installworld' test on -STABLE. > Just tried -j32 installworld with modified install(1) binary for which -S is the default, and it still choked on lib/libncurses. The problem is in bsd.man.mk: we should separate installing of manpages and their MLINKS, otherwise links could be attempted before their originals are installed. Continuing with -DNOMAN now... Cheers, -- Ruslan Ermilov Oracle Developer/DBA, ru@sunbay.com Sunbay Software AG, ru@FreeBSD.org FreeBSD committer, +380.652.512.251 Simferopol, Ukraine http://www.FreeBSD.org The Power To Serve http://www.oracle.com Enabling The Information Age To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010420204116.A60253>