This exp-run request came from bug 202516, the exp-run request for CMake 3.3.1. Patch: https://people.freebsd.org/~rakuco/exp-runs/iconv.mk-cmake-vars-v1.patch There were a lot of problems with iconv in that exp-run, and Tijl and I have been discussing how to address it. I would like to solve this issue independently and before continuing with the CMake 3.3 work. I've tested the patch above (modulo the change in converters/libiconv) with several ports that depend on iconv and CMake. Several of them could be simplified as this patch shows. It would be good to check this on 10.1 (the most problematic FreeBSD release in terms of iconv support) as well as either 9.3 or HEAD.
Tijl: please take a look at the patch. I can also either attach it here or upload it to Phabricator for comments. I've tested CMake 3.3.1 with this patch and ports like libstreamanalyzer build fine on 10.1. I can provide a list of ports that I've been testing as well.
The patch looks good to me except maybe this: Index: textproc/ctpp2/Makefile =================================================================== --- textproc/ctpp2/Makefile (revision 395669) +++ textproc/ctpp2/Makefile (working copy) @@ -36,15 +37,11 @@ DISCARD_ILSEQ_CMAKE_OFF= -DICONV_DISCARD_ILSEQ=OFF TRANSLITERATE_CMAKE_ON= -DICONV_TRANSLITERATE=ON TRANSLITERATE_CMAKE_OFF= -DICONV_TRANSLITERATE=OFF +TRANSLITERATE_USES= iconv:wchar_t iconv:translit here?
Thanks for spotting the oversight. I've uploaded https://people.freebsd.org/~rakuco/exp-runs/iconv.mk-cmake-vars-v2.patch with the fix. Antoine: it's fine if the exp-run has already started, as both versions of the patch have the same effect.
I launched exp-run on 9.3/10.1/10.2 On 9.3 there are already a few new failures due to iconv*() removal (exp-run halfway) http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-01_08h28m27s/logs/errors/deadbeef-0.6.2_5.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-01_08h28m27s/logs/errors/hidapi-0.8.0.r1_3.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-01_08h28m27s/logs/errors/vmw-060510.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-01_08h28m27s/logs/errors/ja-chasen-base-2.4.5_1.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-01_08h28m27s/logs/errors/xapian-omega-1.2.21.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-01_08h28m27s/logs/errors/x3270-3.3.15_2.log
Thanks, I was about to bring that up too. They all seem to come from the removal of that patch from converters/libiconv. How about landing that one separately from the rest?
Exp-run results on 10.2 amd64: http://package18.nyi.freebsd.org/build.html?mastername=102amd64-default-PR202798&build=2015-09-02_05h41m56s 2 new failures: + {"origin"=>"irc/weechat", "pkgname"=>"weechat-1.1.1_1", "phase"=>"package", "errortype"=>"???"} + {"origin"=>"irc/weechat-devel", "pkgname"=>"weechat-devel-20140213_5", "phase"=>"package", "errortype"=>"???"} Failure logs: http://package18.nyi.freebsd.org/data/102amd64-default-PR202798/2015-09-02_05h41m56s/logs/errors/weechat-1.1.1_1.log http://package18.nyi.freebsd.org/data/102amd64-default-PR202798/2015-09-02_05h41m56s/logs/errors/weechat-devel-20140213_5.log
Exp-run results on 10.1 i386: http://package22.nyi.freebsd.org/build.html?mastername=101i386-default-PR202798&build=2015-09-02_09h38m43s 0 new failure
Exp-run results on 9.3 i386: http://package23.nyi.freebsd.org/build.html?mastername=93i386-default-PR202798&build=2015-09-02_10h43m42s 17 new failures: + {"origin"=>"audio/deadbeef", "pkgname"=>"deadbeef-0.6.2_5", "phase"=>"build", "errortype"=>"linker_error"} + {"origin"=>"comms/hidapi", "pkgname"=>"hidapi-0.8.0.r1_3", "phase"=>"configure", "errortype"=>"configure_error"} + {"origin"=>"deskutils/fbreader", "pkgname"=>"fbreader-0.99.6_1", "phase"=>"build", "errortype"=>"linker_error"} + {"origin"=>"deskutils/ljclive", "pkgname"=>"ljclive-0.4.10_1", "phase"=>"build", "errortype"=>"missing_header"} + {"origin"=>"deskutils/owncloudclient", "pkgname"=>"owncloudclient-1.8.4", "phase"=>"build", "errortype"=>"compiler_error"} + {"origin"=>"emulators/vmw", "pkgname"=>"vmw-060510", "phase"=>"build", "errortype"=>"linker_error"} + {"origin"=>"irc/scrollz", "pkgname"=>"scrollz-2.2.3_1", "phase"=>"build", "errortype"=>"clang"} + {"origin"=>"japanese/chasen-base", "pkgname"=>"ja-chasen-base-2.4.5_1", "phase"=>"configure", "errortype"=>"configure_error"} + {"origin"=>"japanese/uim-mozc", "pkgname"=>"ja-uim-mozc-2.17.2106.102", "phase"=>"build", "errortype"=>"missing_header"} + {"origin"=>"java/jboss71", "pkgname"=>"jboss71-7.1.3", "phase"=>"build", "errortype"=>"???"} + {"origin"=>"net/c3270", "pkgname"=>"c3270-3.3.14ga11_1", "phase"=>"configure", "errortype"=>"configure_error"} + {"origin"=>"www/htmlcxx", "pkgname"=>"htmlcxx-0.85_2", "phase"=>"build", "errortype"=>"linker_error"} + {"origin"=>"www/httrack", "pkgname"=>"httrack-3.48.21_1", "phase"=>"build", "errortype"=>"linker_error"} + {"origin"=>"www/p5-Gtk2-WebKit", "pkgname"=>"p5-Gtk2-WebKit-0.09_2", "phase"=>"build", "errortype"=>"???"} + {"origin"=>"www/xapian-omega", "pkgname"=>"xapian-omega-1.2.21", "phase"=>"configure", "errortype"=>"configure_error"} + {"origin"=>"x11/mrxvt-devel", "pkgname"=>"mrxvt-devel-0.5.4_10", "phase"=>"build", "errortype"=>"linker_error"} + {"origin"=>"x11/x3270", "pkgname"=>"x3270-3.3.15_2", "phase"=>"configure", "errortype"=>"configure_error"} Failure logs: http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/deadbeef-0.6.2_5.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/hidapi-0.8.0.r1_3.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/fbreader-0.99.6_1.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/ljclive-0.4.10_1.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/owncloudclient-1.8.4.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/vmw-060510.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/scrollz-2.2.3_1.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/ja-chasen-base-2.4.5_1.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/ja-uim-mozc-2.17.2106.102.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/jboss71-7.1.3.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/c3270-3.3.14ga11_1.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/htmlcxx-0.85_2.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/httrack-3.48.21_1.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/p5-Gtk2-WebKit-0.09_2.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/xapian-omega-1.2.21.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/mrxvt-devel-0.5.4_10.log http://package23.nyi.freebsd.org/data/93i386-default-PR202798/2015-09-02_10h43m42s/logs/errors/x3270-3.3.15_2.log
(In reply to Raphael Kubo da Costa from comment #5) I'll fix the FreeBSD 9 failures and open a separate bug for that.
OK, so shall I remove that part of the change from this patch. If that's the case, there should be only irc/weechat and irc/weechat-devel to fix (I have a fix ready).
(In reply to Raphael Kubo da Costa from comment #10) Yes, go ahead. All changes to converters/libiconv and the BUILD_DEPENDS change in Mk/Uses/iconv.mk can be removed.
Created attachment 160634 [details] Proposed patch Hopefully final version attached to the bug and also available here: https://people.freebsd.org/~rakuco/exp-runs/iconv.mk-cmake-vars-v3.patch It drops the changes to converters/libiconv and fixes irc/weechat and irc/weechat-devel. The rest is identical. Antoine: can I get a portmgr signoff?
I will test this and let you know tomorrow.
Looks good, approved.
A commit references this bug: Author: rakuco Date: Thu Sep 3 15:44:17 UTC 2015 New revision: 395972 URL: https://svnweb.freebsd.org/changeset/ports/395972 Log: Uses/iconv.mk: Set iconv-related CMake variables. The way we deal with iconv in base and ports across different FreeBSD releases is complicated: 9.x does not have iconv.h in base, 10.1 has it with a different prototype for iconv(3) and later versions have the right iconv(3) prototype. And, in some cases (USES=iconv:{translit,wchar_t}), we must always use the libiconv port. This is why there are so many checks in Uses/iconv.mk: we need to know the situation we currently have in order to decide whether to pull iconv from converters/libiconv, whether to just use its header (and pull the library from base) or whether to use everything from base. r384038 adjusted several CMake-based ports, but did so in a way that was not very scalable and required a few intrusive patches to some ports. Most ports that have both USES=cmake and USES=iconv use variations of FindIconv.cmake that behave similarly. This change passes the header and library values we really want to use to CMake using the most common variable names, bypassing the calls to find_path() and find_library() that would sometimes end up finding the wrong file. The few ports that use different variable names have had their Makefiles adjusted (we manually pass the values we want via CMAKE_ARGS). Other changes: - chinese/fcitx: Explicitly set LIBICONV_LIBC_HAS_ICONV_OPEN=OFF as we always want the version from ports because of USES=iconv:wchar_t. - editors/calligra: Explicitly use iconv:translit because Kexi needs it. - irc/weechat and irc/weechat-devel: The FindIconv.cmake patches could not be entirely removed because the check_library_exists() calls are wrong. Sent upstream: https://github.com/weechat/weechat/pull/513 - textproc/ctpp2: Use iconv:translit when the TRANSLITERATE option is used. PORTREVISION has been bumped in editors/calligra and textproc/ctpp2 because their dependency list has changed in 10.2 and later as the ports version is always used now. PR: 202798 Reviewed by: antoine, tijl Approved by: portmgr (antoine) Changes: head/Mk/Uses/iconv.mk head/audio/tagutil/files/ head/chinese/fcitx/Makefile head/chinese/fcitx/files/patch-cmake__FindLibiconv.cmake head/deskutils/libstreamanalyzer/Makefile head/deskutils/libstreams/Makefile head/editors/calligra/Makefile head/irc/weechat/Makefile head/irc/weechat/files/patch-cmake-FindIconv.cmake head/irc/weechat-devel/Makefile head/irc/weechat-devel/files/patch-cmake-FindIconv.cmake head/mail/libvmime/files/patch-cmake__FindIconv.cmake head/net-im/licq/files/patch-cmake-Modules-FindIconv.cmake head/textproc/ctpp2/Makefile head/textproc/ctpp2/files/patch-CMakeLists.txt head/textproc/simplexml/Makefile head/textproc/simplexml/files/patch-cmake__FindIconv.cmake head/textproc/wbxml2/files/patch-cmake-modules-FindIconv.cmake
Landed. There's a slight difference between v3 and the version I committed: I've bumped editor/calligra's and textproc/ctpp2's PORTREVISIONs following the changes from USES=iconv to USES=iconv:translit which always requires the ports version. Thanks everyone!