Date: Fri, 1 Nov 2002 13:25:39 -0800 From: Marcel Moolenaar <marcel@xcllnt.net> To: Steve Kargl <sgk@troutmask.apl.washington.edu> Cc: Daniel Flickinger <attila@hun.org>, FreeBSD-CURRENT <current@FreeBSD.ORG> Subject: Re: another include failure to find in buildworld Message-ID: <20021101132539.B85578@kayak.xcllnt.net> In-Reply-To: <20021101205933.GA23809@troutmask.apl.washington.edu> References: <20021101072304.sAZL15425@hun.org> <20021101125038.A85578@kayak.xcllnt.net> <20021101205933.GA23809@troutmask.apl.washington.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Nov 01, 2002 at 12:59:33PM -0800, Steve Kargl wrote: > On Fri, Nov 01, 2002 at 12:50:38PM -0800, Marcel Moolenaar wrote: > > On Fri, Nov 01, 2002 at 07:23:04AM +0000, Daniel Flickinger wrote: > > > > > > This is another instance where the build is not reading > > > from the /usr/obj tree, reading from /usr/include first. > > > > I don't think so. You cannot do cross-builds if you're messing > > up the include searches. I would suggest you check out a clean > > source tree, revert /usr/include to a state where you know it > > failed before (ie remove /usr/include/uuid.h for example) and > > start a *non-parallel* build without any options like -k or -s > > for target buildworld *AFTER* validating and preferrably nuking > > /etc/make/.conf. There's really no point complaining on the list > > about breakages that you only see. If the failure is real, we at > > least need to be able to reproduce it before we can fix it and > > so far you're the only one with problems. > > > > I'll do the same to make sure my claim that you're the only one > > who sees this has been verified for me for the latest sources... > > > > Don't waste your time, Marcel. Unless Daniel has changed his > build procedure, he uses a custom script to do the builds. See > > http://www.freebsd.org/cgi/getmsg.cgi?fetch=491021+500131+/usr/local/www/db/text/2002/freebsd-current/20021020.freebsd-current Thanks Steve! That's vital information... The buildworld script is broken on two accounts: 1. it does an installincludes after completely cleaning the object tree and thus is very likely failing because buildincludes is not run. This may be hidden by -k. 2. it's obviously not delivering on its primary objective of reducing the marging of error. In fact, it may even contribute to the error because of the excessive use of -k and redirecting to /dev/null. Adding NO_WERROR by default seems to defeat the purpose as well. In any event, one would want two successive 'make cleandir' runs if one wants to be sure everything is as clean as possible. Other than that, I don't see why the script itself is causing the make buildworld to fail. It increases the chance that the source tree is borked or deliberately modified (which may be equivalent statements :-) -- Marcel Moolenaar USPA: A-39004 marcel@xcllnt.net To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20021101132539.B85578>