Date: Fri, 26 Aug 2011 20:02:34 +0000 From: Alexey Dokuchaev <danfe@FreeBSD.org> To: Koop Mast <kwm@FreeBSD.org> Cc: cvs-ports@FreeBSD.org, cvs-all@FreeBSD.org, ports-committers@FreeBSD.org Subject: Re: cvs commit: ports/textproc/ots Makefile Message-ID: <20110826200234.GA8743@FreeBSD.org> In-Reply-To: <201108261828.p7QISje3085996@repoman.freebsd.org> References: <201108261828.p7QISje3085996@repoman.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Aug 26, 2011 at 06:28:45PM +0000, Koop Mast wrote: > kwm 2011-08-26 18:28:45 UTC > > Modified files: > textproc/ots Makefile > Log: > Mark MAKE_JOBS_UNSAFE. Just replying to a random MAKE_JOBS_UNSAFE commit. I have a suggestion which I think would help us to fix parallel builds when portmgr@ decides to start running -exp runs with forced MAKE_JOBS: every time one of us marks a port as MAKE_JOBS_UNSAFE, should provide a short comment on the nature of a failure. While in majority of cases it can hardly be generalized as it is probably due to the incorrectly defined dependencies, sometimes it's one of the well-known cases: missing `-p' switch to mkdir(1) calls, GNU make vs. BSD make (devel/nasm is an example of this particular case), or if it only fails for -j4 but not -j2. Perhaps MAKE_JOBS_UNSAFE knob could be refactored into something similar to other "brokenness" knobs and tought to accept a reason string, while "yes" value could still be used for cases when exact cause of breakage is not so obvious. I also encourage everyone who experience brokenness during parallel builds check what popular GNU/Linux distros think about it. AFAIK, Gentoo does parallel builds by default, and explicitly passes -j1 in their ebuilds if they don't have relevant patch. Sometimes the patch exists, in which case it should be backported to FreeBSD. Ditto for OpenBSD, Debian, RedHat... ./danfe
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20110826200234.GA8743>