Date: Wed, 30 Oct 2002 14:04:44 -0800 From: Marcel Moolenaar <marcel@xcllnt.net> To: Daniel Flickinger <attila@hun.org> Cc: Hiten Pandya <hiten@angelica.unixdaemons.com>, FreeBSD-CURRENT <current@FreeBSD.ORG> Subject: Re: libc in CURRENT fails as of 1200 GMT today Message-ID: <20021030140444.A79706@kayak.xcllnt.net> In-Reply-To: <20021030205505.FwhS20265@hun.org> References: <20021030164823.NvQM9234@hun.org> <20021030120635.A21427@angelica.unixdaemons.com> <20021030205505.FwhS20265@hun.org>
next in thread | previous in thread | raw e-mail | index | archive | help
[The crucial question is hidden somewhere...] On Wed, Oct 30, 2002 at 08:55:05PM +0000, Daniel Flickinger wrote: > Sent: Wed, 30 Oct 2002 09:02:17 -0800 by Marcel Moolenaar > > On Wed, Oct 30, 2002 at 04:48:23PM +0000, Daniel Flickinger wrote: > + > /usr/src/lib/libc/uuid/uuid_compare.c:31:18: uuid.h: No such file or directory > + > [snip] > + > > + > and a couple pages of resulting syntax errors > + > > + > + I assume you didn't do a makeworld, but just a make in libc? > > No, I go for broke... Why doubt if you can be sure :-) > I 'buildworld' at least once a day based on 1200 GMT > --other times, if necessary. The system is CURRENT as of > 1800 GMT 27 Oct 2002, both kernel and world. The last two > days have had too many errors to go for installworld. I buildworld alpha, i386 and ia64 yesterday without any problems. I did an installworld for alpha and ia64 without problems. The i386 installworld is going to happen once I'm home. The point: I have no idea what problems you're referring to other than the typical "commit->breakage->fix" cycles we have. > find /usr -name 'uuid\.h' -print > > /usr/obj/usr/src/i386/usr/include/uuid.h > /usr/include/sys/uuid.h > /usr/src/lib/libc/uuid/uuid.h > /usr/src/sys/sys/uuid.h This list is correct. Note that /usr/obj/usr/src/i386/usr/include/sys is a link to /usr/src/sys/sys. This explains why you don't see sys/uuid.h under /usr/obj/usr/src/i386/usr/include, but it's visible there alright. > However, after running the script above we now have a > uuid.h in both /usr/include and /usr/include/sys: > > find /usr -name 'uuid\.h' -print > > /usr/obj/usr/src/i386/usr/include/uuid.h > /usr/include/sys/uuid.h > /usr/include/uuid.h > /usr/src/lib/libc/uuid/uuid.h > /usr/src/sys/sys/uuid.h However? This is correct again. > The include/uuid.h reads in the sys/uuid.h file. This now > provides a fallback, masking the real problem. [here it is] Could you explain to me what you think the real problem actually is? So far you described correct behaviour and labeled it as broken, so you have to excuse me if I prefer to assume as little as possible now. > Both files are present in the src tree --which is where > 'make buildworld' SHOULD find them and place them in the > /usr/obj tree during buildworld. There is a missing -I > declaration ... or cp or install statement. It is finding both in the object dir. At least on my systems. I specifically tested this by removing /usr/include/uuid.h prior to starting a buildworld that didn't had -DNOCLEAN. > It SHOULD NOT be necessary to run 'make > build/installincludes' prior to 'make buildworld' Correct and that's what it currently is: not necessary. > After 23 years of BSD, it's just another day... Happiness devaluates just as money does. After experiencing the joy of BSD for 23 years I'm not surprised that you've grown not to notice it anymore and thus that it feels like just another day. Try Windows for an hour and you'll feel alive and refreshed again... :-) -- 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?20021030140444.A79706>