Re: make release fails
On 05/23/2013 06:52 AM, Jack Mc Lauren wrote: Hi snip find //usr/obj/usr/src/release/dist/doc -empty -delete find: -delete: //usr/obj/usr/src/release/dist/doc: relative path potentially not safe *** [distributeworld] Error code 1 snip What's wrong with this? Thanks in advance Huh. Apparently I ran across this myself in the past, since I found a patch lurking in my source tree. 8 diff --git a/Makefile.inc1 b/Makefile.inc1 index 4567e5d..1830483 100644 --- a/Makefile.inc1 +++ b/Makefile.inc1 @@ -685,7 +685,7 @@ distributeworld installworld: installcheck ${IMAKEENV} rm -rf ${INSTALLTMP} .if make(distributeworld) .for dist in ${EXTRA_DISTRIBUTIONS} - find ${DESTDIR}/${DISTDIR}/${dist} -empty -delete + find ${DESTDIR}/${DISTDIR}/${dist} -empty -exec rmdir {} + .endfor .endif 8 The reason this occurs is because you and I are building with NODOC, which leaves dist/doc empty, and the above find construct will refuse to -delete if the directory specified on the command line is one of the ones that would have been deleted. Either patch the makefile as above, tell make to ignore the return code of this find invocation, or put some non-empty files into /usr/obj/usr/src/release/dist/doc during the release building process (like a README pointing to the docs tarball on the mirrors) to make it not trigger the empty condition. Hope this helps! -- Fuzzy love, -CyberLeo Technical Administrator CyberLeo.Net Webhosting http://www.CyberLeo.Net cyber...@cyberleo.net Furry Peace! - http://www.fur.com/peace/ ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
make release fails
Hi I'm trying to make my own release ... # cd /usr/src # make buildkernel KERNCONF=MYKERN # make -j4 buildworld # cd release/ # make release NODOC=YES NOPORTS=YES NOSRC=YES and that's the error while making release: find //usr/obj/usr/src/release/dist/doc -empty -delete find: -delete: //usr/obj/usr/src/release/dist/doc: relative path potentially not safe *** [distributeworld] Error code 1 Stop in /usr/src. *** [distributeworld] Error code 1 Stop in /usr/src. *** [base.txz] Error code 1 Stop in /usr/src/release. *** [release] Error code 1 Stop in /usr/src/release These are contents of /etc/src.conf file : WITHOUT_BLUETOOTH=YES WITHOUT_EXAMPLES=YES WITHOUT_FLOPPY=YES WITHOUT_GAMES=YES WITHOUT_MAN=YES WITHOUT_MAN_UTILS=YES WITHOUT_SHAREDOCS=YES WITHOUT_WIRELESS=YES WITHOUT_WIRELESS_SUPPORT=YES WITHOUT_AT=YES WITHOUT_CALENDAR=YES WITHOUT_INFO=YES WITHOUT_LOCALES=YES WITHOUT_ZFS=YES WITHOUT_BSD_CPIO=YES WITHOUT_CTM=YES WITHOUT_DICT=YES WITHOUT_GDB=YES WITHOUT_GNU=YES WITHOUT_GROFF=YES WITHOUT_HTML=YES WITHOU_INFO=YES WITHOUT_LPR=YES WITHOUT_MAIL=YES WITHOUT_PORTSNAP=YES WITHOUT_QUOTAS=YES WITHOUT_RCS=YES WITHOUT_SYSINSTALL=YES WITHOUT_BIND=YES WITHOUT_BIND_XML=YES WITHOUT_BIND_IDN=YES WITHOUT_BIND_SIGCHASE=YES WITHOUT_BIND_LARGE_FILE=YES WITHOUT_FREEBSD_UPDATE=YES WITHOUT_RESCUE=YES What's wrong with this? Thanks in advance ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Follow-up: Yet Another make release fails on ghostscript-gnu
Michael Dexter [EMAIL PROTECTED] writes: Having complete control over the build of my network OS is simply revolutionary... but I was hoping this revolution would not be so bloody. You can get *that* with one of the supported update options. make release was never intended for anybody but release engineers. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Yet Another make release fails on ghostscript-gnu
Greetings from Latvia, I am very, very excited about FreeBSD's make release ability but like many others I have read about on the various list archives, I my builds are hanging on trivial failures. In this case, the eplaser-3.0.4-651.tgz file is failing its checksum, bringing the build to a halt. I have followed the advice of adding NO_CHECKSUM=yes to no avail. I have tried it both after make and at the end of my string of variables. Which reads: make release CHROOTDIR=/usr/testrelease BUILDNAME=4.7-RELEASE CVSROOT=/usr/ncvs RELEASETAG=RELENG_4_7_0_RELEASE NOPORTS=yes NOPORTREADMES=yes NODOCS=yes TARGET_ARCH=i386 TARGET=i386 (/usr/ncvs is my choice, NO_CHECKSUMS=yes has again gone either after make or at the end) I have also: 1. Downloaded the same file from other /usr/ports/distfiles directories. Though different sizes, they also fail the check. 2. Tried to fetch new versions from /usr/ports/print/ghostscript-gnu/ and place them in either /usr/ports/distfiles and /usr/ports/ghostscript/ AND /usr/ports/ghostscript-gnu/ 3. Not tried DOMINIMALDOCPORTS as this would seem to go through the minimum files I am already having trouble with. Observations: 1. NO_CHECKSUM does not appear to be in any of the Makefiles in /usr/src/release/ or /usr/src/ Where is this check being called? Is there any other way to override it? Does it really exist? 2. The only vague reference to the file I need is in /usr/src/release/Makefile.inc.docports and reportedly the ports. man ports reveals the elusive NO_CHECKSUMS option! I see /usr/src/release/Makefile, doc and scripts but alas my make is not fluent enough to understand who is calling what and when. I would hope I do not need to feed NO_CHECKSUMS into evey invocation of make. 3. NOPORTSATALL=yes seems to be like an unavailable option, given that there will always be minimum documentation, which appear to need ports. 4. Others in need have pointed out that a build-friendly distfiles tar ball would ease this process. Does such a thing exist? Could I strip/re-create it from the official 4.7 release? Note that I have brought my sys, ncvs and ports source trees up to date. Because things are downloaded, I don't see how a fresh, raw, official 4.7 install would help what seems to be the checksum of a single downloaded port that I understand is downloaded on demand. THIS bothers me: make checksum inside of /usr/ports/print/ghostscript-gnu PASSES ALL CHECKSUM TESTS, including eplaser-3.0.4-651.tgz Why would this fail during make release? Any ideas? Many thanks, Michael. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Follow-up: Yet Another make release fails on ghostscript-gnu
Greetings again, Regarding the make release mentioned earlier... After peppering my /usr/ports/distfiles/ tree with very redundant distfiles, i.e. the same files in /usr/ports and /usr/ports/ghostscript and /usr/ports/ghostscript-gnu, things finally worked. (Ending on a vn present failure but I know where to look on that one) Given the time it takes to test each and every variable, I fear I will never be confident that I have an answer, though it was very educational... Early on, I tried stepping through make release.1 and all but this appeared to ignore the flags I sent it. All of the output was sent to the /R directory, rather than my choice of /usr/testrelease/ Can that be changed? Is there indeed a way to step-trough a release build? Having complete control over the build of my network OS is simply revolutionary... but I was hoping this revolution would not be so bloody. Conclusions: As suggested elsewhere, it would be nice to have an official source of buildable release files like the /usr/src on the CD, that would spare one the guesswork of trusting the ports fetch to build the build files prior to making the release. Might a pre-make script based upon the real make script perform all of the downloads and checksum verifications? This could save hours in wasted build time and guesswork, a bit like running cvsup prior to building world, knowing exactly what source will be used. I will look into this but again, my experience with make is only a few hours old. Thanks again, Michael. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message