Created attachment 144846 [details] proposed patch The buildprocess complains about: Making all in doc make[4]: make[4]: don't know how to make protocol-binary.txt. Stop It seems to need "USES= gmake" While at it, I also made portlint (nearly) happy, see attached diff
I'm unable to reproduce any issue with either option enabled. Can you perhaps show the log file for the build where you are having issues?
Created attachment 144862 [details] typescript containing build log See my comment mentioning this attachment
(In reply to Steve Wills from comment #1) > I'm unable to reproduce any issue with either option enabled. Can you > perhaps show the log file for the build where you are having issues? See the attachment "typescript" I also tried to build the otiginal under redports with disastrous results, see as example: https://redports.org/~jaap/20140721194500-62439-226258/.txz.log
Ok, from the log, this seems to be happening because textproc/xml2rfc and textproc/libxslt are installed, perhaps this is changing which docs are built or how they are built. I will look into it. Thanks!
A commit references this bug: Author: swills Date: Tue Jul 22 19:30:32 UTC 2014 New revision: 362570 URL: http://svnweb.freebsd.org/changeset/ports/362570 Log: databases/memcached: fix build when xml2rfc is installed The docs for require xml2rfc and aren't built in the case where it's not installed. However when they are installed, they fail due to make issues. This patches the doc Makefile to avoid the gmake specific extensions as well as ensuring that network access is not needed to build the docs. It also ensure that xml2rfc is always there in the case where the DOCS option is enabled so that the docs are always built the same. While here, I decided to include some of the cleanup from the PR including the LICENSE change ane enabling the SASL and DOCS options by default since they don't add much in the way of dependencies. Also switched to static plist. PR: 192009 Reported by: jaap@NLnetLabs.nl Changes: head/databases/memcached/Makefile head/databases/memcached/files/patch-doc-local head/databases/memcached/files/patch-doc__Makefile.in head/databases/memcached/pkg-plist
Took most of the changes from this patch along with a few others to clean some things up, see commit message. I think portlint lead you astray on the distdir thing, the heuristics there aren't perfect and I think the distdir thing was incorrect. Please let me know if you have any further issues, but I believe this should resolve it. Thanks!
(In reply to Steve Wills from comment #6) > Took most of the changes from this patch along with a few others to clean > some things up, see commit message. I think portlint lead you astray on the > distdir thing, the heuristics there aren't perfect and I think the distdir > thing was incorrect. Please let me know if you have any further issues, but > I believe this should resolve it. Thanks! This indeed fixes building the port. However, if you want to build this in a tinderbox (I have my own,but resorts.org also shows it) it wants to build all ports in /usr/ports/lang and dies after a while. I do think this is trigger by the TEST_DEPENDS line, expecially since portlints says: WARN: Makefile: no port directory /usr/ports/lang/${PERL_PORT} found, even though it is listed in TEST_DEPENDS. Anyway, for just using it with as a port it works fine. Thanks! jaap
(In reply to jaap from comment #7) > (In reply to Steve Wills from comment #6) > > Took most of the changes from this patch along with a few others to clean > > some things up, see commit message. I think portlint lead you astray on the > > distdir thing, the heuristics there aren't perfect and I think the distdir > > thing was incorrect. Please let me know if you have any further issues, but > > I believe this should resolve it. Thanks! > > This indeed fixes building the port. However, if you want to build this in a > tinderbox (I have my own,but resorts.org also shows it) it wants to build > all ports in /usr/ports/lang and dies after a while. I do think this is > trigger by the TEST_DEPENDS line, expecially since portlints says: > > WARN: Makefile: no port directory /usr/ports/lang/${PERL_PORT} found, > even though it is listed in TEST_DEPENDS. > > Anyway, for just using it with as a port it works fine. Thanks! > > jaap Went ahead and removed the TEST_DEPENDS for now, we can add it back later.
(In reply to Steve Wills from comment #8) > (In reply to jaap from comment #7) > > (In reply to Steve Wills from comment #6) > > > Took most of the changes from this patch along with a few others to clean > > > some things up, see commit message. I think portlint lead you astray on the > > > distdir thing, the heuristics there aren't perfect and I think the distdir > > > thing was incorrect. Please let me know if you have any further issues, but > > > I believe this should resolve it. Thanks! > > > > This indeed fixes building the port. However, if you want to build this in a > > tinderbox (I have my own,but resorts.org also shows it) it wants to build > > all ports in /usr/ports/lang and dies after a while. I do think this is > > trigger by the TEST_DEPENDS line, expecially since portlints says: > > > > WARN: Makefile: no port directory /usr/ports/lang/${PERL_PORT} found, > > even though it is listed in TEST_DEPENDS. > > > > Anyway, for just using it with as a port it works fine. Thanks! > > > > jaap > > Went ahead and removed the TEST_DEPENDS for now, we can add it back later. Yes, taking out the TEST_DEPENDS takes enabling building in tinderboxes and on redports.org. Thanks again.