Why clang++37 behaves differently on 9.3 and 10.3?
clang++37 compiles this simple program fine on 10.3, but fails on 9.3. Why does it behave differently on different OS versions? It looks like it ignores -std=c++11 on 9.3. Yuri ---program--- #include int main() { const int vmax = std::numeric_limits::max(); static_assert(vmax>0, ""); } ---command--- clang++37 -std=c++11 -c test.cc ---error on 9.3--- test.cc:6:17: error: static_assert expression is not an integral constant expression static_assert(vmax>0, ""); ^~ test.cc:6:17: note: initializer of 'vmax' is not a constant expression test.cc:5:13: note: declared here const int vmax = std::numeric_limits::max(); ^ 1 error generated. ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Please commit this maintainer times out patch: Bug 199364 - emulators/virtualbox-ose: Fix for the KDE dialog problem (edit)
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=199364 It fixes the long standing problem of file dialog in VirtualBox UI under kde. Thanks, Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Please commit the new ports databases/sqlite-ext-regexp and databases/sqlite-ext-spellfix - they are pending from Sep 2015
These are SQLite DB extensions. These are very simple ports. Somebody might need them. https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=203217 https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=203218 I just updated them and made sure they build. Thanks! Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: Clang can't build c++ project: ubsan is missing
On 02/26/2016 13:50, Dimitry Andric wrote: It looks like you are building the port with -fsanitize=undefined, and your clang installation(s) are missing the UBSan libraries, for some reason. Try building your port without -fsanitize=undefined, unless you need that for something specific, of course. Thanks! It turned out that sanitizer is the relatively new feature that isn't yet available in any of the clang versions on FreeBSD, and this package fails to detect the compiler support for it. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Clang can't build c++ project: ubsan is missing
Hi, I can't update the version of one C++ port because now the base clang-34 fails to find the symbols __ubsan_ (for example __ubsan_vptr_type_cache). clang36 port compiler fails to find libclang_rt.san-x86_64.a clang38 port compiler fails to find libclang_rt.ubsan_standalone-x86_64.a What could be wrong? Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: When is Qt-5.5 coming to ports?
Since it turned out that nobody was working on this, I ported qt-5.5.1 myself. Looking of a committer to commit this patch: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=205805 Thanks, Yuri On 12/10/2015 16:18, Yuri wrote: Is it coming any time soon? It blocks several ports. Many Qt-based apps now want 5.5. ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: USES=fortran can't mix with the libraries requiring /lib/libgcc_s.so.1 from the base
On 12/24/2015 18:23, Shane Ambler wrote: Another possibility might be DragonEgg, though it doesn't appear to have been updated for a while. http://dragonegg.llvm.org/ I tried DragonEgg, but it is only the optimization/code generation plugin to gcc, and gcc still writes the final binaries the same way. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: USES=fortran can't mix with the libraries requiring /lib/libgcc_s.so.1 from the base
On 12/23/2015 09:23, Diane Bruce wrote: > >No. The core problem is due to our version of libgcc not having quadmath > >support. > > is not that simple. Please read the various threads to understand why. What does it take to update the base libgcc to the version with quadmath support? This is a very bad problem, because it prevents most science and math software from running on FreeBSD. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: USES=fortran can't mix with the libraries requiring /lib/libgcc_s.so.1 from the base
On 12/23/2015 06:34, Diane Bruce wrote: No. The core problem is due to our version of libgcc not having quadmath support. If the separate port would have been created for gcc with only fortran in it, and it would have been compiled with clang, this would have solved this problem. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
USES=fortran can't mix with the libraries requiring /lib/libgcc_s.so.1 from the base
I found that ports with USES=fortran can't mix with anything in C++ compiled with the base clang++, because USES=fortran forces the current gcc that links with its /usr/local/lib/libgcc_s.so.1 Getting this particular error from the python process, because one python module has USES=fortran, and another has C++ in it: ImportError: /lib/libgcc_s.so.1: version GCC_4.6.0 required by /usr/local/lib/gcc48/libgfortran.so.3 not found What is the general solution for this problem? Is there a non-gcc version of fortran? One thing is when gcc is required because clang can't compile something, and another things is when fortran language requires it. The latter is here to stay. Can there be the separate fortran from gcc that is build with clang? Or can we switch /usr/ports/lang/gccNN to be always built with the base clang? I know this is certainly possible. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: Gogs port: A painless self-hosted Git service
On 12/18/2015 19:44, Craig Rodrigues wrote: Would you be able to help test this script which takes a patch and then kicks off poudriere: https://github.com/Ultima1252/portest/blob/master/portest and provide feedback or patches? Ricky Gallamore wrote it. This type of script will be necessary to implement the kind of workflow which you are suggesting. Integrating with pull requests is doable down the road, but we need to get the basic scripting working at first. I am working with gogs authors to have all needed features implemented. Currently gogs still doesn't have the hook for the pull request creation, which is a must. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: Gogs port: A painless self-hosted Git service
On 12/13/2015 23:54, Torsten Zühlsdorff wrote: I did this often for different processes. I won't work on the issue this year - there is do much open to do. But please write me a desired workflow and the needed requirements and i will do this in the next year. Just 3 weeks to go :D This isn't a complete workflow, but here is an idea for port updates: Each port update or new port submission could become a pull request. Once the new pull request is created, system should launch poudriere builds for it for all required platforms, and append the notice that builds have started. Then append the build status. If it was successful, committer can just approve the change with one press of a button. I believe there is a way to do a two way sync between git and subversion. If at any point in time that port changes, the process has to be invalidated and restarted if the pull request is still mergeable, or submitter should be notified if it became unmergeable. This could simplify and expedite the process of committing of port changes. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: poudriere, Go and networking
On 12/11/2015 05:57, Piotr Florczyk wrote: Recently I had to package couple of programs written in Go and godep is becoming the standard for dependency tracking in Go projects. For example I currently had to package telegraf. Here is the thing. Poudriere disables networking after fetch phase and I don't know before extract phase what dependencies are inside. You don't need networking in extract phase. I recently made go.mk and nodejs.mk - both manage to get dependencies w/out networking: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=205282 and https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=204577 It is *extremely dangerous* (though convenient) to let dependencies download automatically, like Go and NodeJS do. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: Gogs port: A painless self-hosted Git service
On 12/13/2015 03:49, Matthew Seaman wrote: The big advantage of being able to use pull requests on Github is that so many people who work on open source projects are already setup to use Github. They've already jumped through all the hoops necessary for them to be able to easily clone a repo, develop some changes and submit them back to the original project. Having a low barrier to entry like that is a really big deal. That's the thing: it looks just as github itself, only the host name in URL is different. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Gogs port: A painless self-hosted Git service
Some time back it was a discussion on this ML about how to best submit patches to FreeBSD ports and base system. One of the problems standing in the way of improvements was the lack of open source web software to manage public submissions. This problem is now addressed. I created a port for Gogs: "A painless self-hosted Git service". It is the open source equivalent of GitHub, with pretty much similar functionality. https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=205283 Can FreeBSD use it for the pull requests? Maybe using git ⇆subversion convertor? Anyway, Gogs is among the nicest web-based systems I saw. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
When is Qt-5.5 coming to ports?
Is it coming any time soon? It blocks several ports. Many Qt-based apps now want 5.5. Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Is it possible to specify which files are backed up to MASTER_SITE_BACKUP?
I have the complex fetch stage which takes one set of distfiles, and produces another distfile, only that distfile is used in build. There is the variable CKSUMFILES that controls which distfiles are included in distinfo (in my case all of them should be there). But how do I control which distfiles get to MASTER_SITE_BACKUP? Is there a variable that controls this, distinct from DISTFILES? In my case I only want the distfile that my fetch produces to be cached in MASTER_SITE_BACKUP. How can I do this? Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Ports with LOCAL/xxx as a MASTER_SITE
I came across a few ports like this. Most recently www/npm. It usually turns out that some slightly modified version of sources is kept there @ LOCAL/. Why not just have the patches under files/ do the modification, so that it is public and reproducible? There are also GitHub and Bitbucket, among other places, to keep sources. There should be no need to allow LOCAL/xxx as a MASTER_SITE nowadays. I suggest to abolish this practice. Begin with adding a warning to the port infrastructure when LOCAL/xxx is specified. (It makes it more difficult to suggest the patch to such port, because outsiders should first "reverse-engineer" the patches, which should have been there in files/ in the first place.) Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Please make Bug#201540 (tor update to 0.2.6.10 with maintainer timeout) a priority because it fixes the broken 0.2.6.9
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=201540 See this for reference: https://trac.torproject.org/projects/tor/ticket/17059 Yuri ___ freebsd-ports@freebsd.org mailing list https://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"
Re: Any ports that download whole trees with git?
On 06/16/2015 16:33, kpn...@pobox.com wrote: Are there any ports that download entire source trees using git? I'm looking at doing a port, but the only source for the source is a tree in git. Are there any examples I can crib from? I generalized my previous patch into USE_SCMREPO: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=200929 It fetches source from SCM repository. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Any ports that download whole trees with git?
On 06/16/2015 16:33, kpn...@pobox.com wrote: Are there any ports that download entire source trees using git? I'm looking at doing a port, but the only source for the source is a tree in git. Are there any examples I can crib from? Not with git. But some while ago I made a port for gpg4usb which is only distributed in source as the subversion repository. So I made a patch adding USE_SVNREPO to the framework: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=198449 You can take an idea from there. The main point is to make such download reproducible. Times of files/links/etc should be adjusted. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
C++ ports using virtual function deletion (c++11 feature) are broken when clang is used
I discovered yesterday that '__cxa_deleted_virtual' function is missing in libcxxrt exports list, and therefore isn't available. As a consequence, C++ ports using virtual function deletion will likely fail to build. I had to add the copy of this function as a port patch to solve this, but I am not sure what the general solution might be when more ports will hit this problem. Bug report: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=200863 (Added missing '__cxa_deleted_virtual' into libcxxrt) Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: www/firefox really depends on security/openssl?
On 05/13/2015 04:11, Carmel NY wrote: The most reliable method to eliminate this, for lack of a better word bullshit, would be for FreeBSD to keep the base system openssl version up-to-date. It is apparent to even the most casual observer that the present method of allowing to different versions of such an important application on the same system without a fail proof method of choosing which version to use as you have demonstrated is truly counter productive to a stable environment. Even keeping the base up-to-date won't necesarily work, since mixing of two copies of the same shared lib from different locations may, and probably will cause faulty behavior due to static variables, among other reasons. Base OpenSSL should be used for one thigs, and port - for others. Isolation is important. I raised this conversation on Apr 1 here, but apparently this important issue is still not resolved. I can't do this myself, because the patch will be likely touching ~100 places, and people who commit it will have to go through all the details, and essentially redo all the thinking. I can't even get simple and obvious stage-qa checks to be checked in. Likely because they aren't exciting enough. People are attracted to exciting stuff. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: www/firefox really depends on security/openssl?
On 05/12/2015 02:25, Dr. Peter Voigt wrote: Therefore I conclude: - Installing binary packages with pkg does not honor the WITH_OPENSSL_BASE=yes switch. Is there another place to tell pkg to use base openssl when doing binary installations? Binary packages are built with default choices for port options. These choices are fixed, and don't depend on your choice of WITH_OPENSSL_BASE=yes in ports. Also this option WITH_OPENSSL_BASE=yes should be deprecated ASAP in all ports, except maybe very few. - If port openssl is not present on a system, any dependency to openssl is not detected by porttree. OpenSSL is an oddball, because USE_OPENSSL is interpreted in a weird way that it tries to detect its port presence and link with it, so standard packages are often built with base SSL which is a problem. This has been discussed, but I am not sure of when this will be fixed. In short, as I also mentioned before, you won't be able to get rid of OpenSSL port because some packages require it unconditionally. So the best strategy is to use OpenSSL port for everything. You will likely be successful if you build them yourself from ports, and fix places where base SSL comes into play. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: www/firefox really depends on security/openssl?
On 05/11/2015 11:21, Dr. Peter Voigt wrote: Thanks for your feedback. I have to admit that I am a bit lost with the referenced PR: Or you can just run 'pkg info -d firefox' - it doesn't show any openssl dependencies. Running 'ldd' on elfs in firefox package also don't show openssl. So firefox doesn't depend on OpenSSL. Also, you shouldn't be switching to the base OpenSSL just for one package. The reason is, again, the conflict between the base and port OpenSSL versions. If I were you, I would move in another direction, and try to eliminate base OpenSSL dependency. You need to figure out which dependency causes it. You should identify it and see if this can be fixed. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: www/firefox really depends on security/openssl?
On 05/11/2015 17:28, Dr. Peter Voigt wrote: What is wrong with my system? Why does porttree need port openssl as well. It is obviosuly wrong. If I build ports-mgmt/porttree from source, there is not dependency from port openssl. Might there be something wrong with my package database? How can I fix? Python depends on openssl. # porttree ports-mgmt/porttree Depending on 0 +ports-mgmt/porttree 1 ++lang/python27 2 |++devel/libffi 3 ||`print/indexinfo 4 |+devel/ncurses 5 |+devel/pkgconf 6 |`+security/openssl 7 | `lang/perl5.16 8 `+math/py-networkx 9++devel/py-setuptools27 |`+lang/python27 - 1 `+lang/python27 - 1 You won't be able to remove port openssl. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: www/firefox really depends on security/openssl?
On 05/11/2015 15:02, Dr. Peter Voigt wrote: I changed to base openssl after fighting with latest port openssl. As reported by several people https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=198788 You can also downgrade security/openssl port locally to the last known version, and then rebuild postfix, and the rest, locally without packages. Local builds currently generally produce better results compared to poudriere builds because they are more likely to avoid openssl problem. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Update of qbittorrent to 3.2.0 and libtorrent-rasterbar to 1.0.4
Please commit update for these two ports: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=200108 Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: www/firefox really depends on security/openssl?
On 05/09/2015 03:56, Dr. Peter Voigt wrote: I am wondering, if www/firefox really depends on security/openssl. According to the following, it should not: This stage-qa test that I submitted half year ago tests if some package really depends on other packages, and recommends additions to *_DEPENDS statements: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=195203 So if it will say that USE_OPENSSL=yes is needed, this means FF does depend on it. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Poudriere build fails on several ports with leftovers message for info/dir
On 04/13/2015 22:55, Yuri wrote: Here is the message: Error: Files or directories left over: info/dir Followup to this: it turns out that .info files placed in plist (instead of INFO=) cause this. I just developed the new stage-qa test 'infoplist': https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=199978 If you see any .info files in plists, or you see this stage-qa test failing, please correct the corresponding individual ports, I could find 101 such pkg-list files. Yuri ---list of pkg-plist files having .info in them--- # find . -name pkg-plist -and -exec grep -l \.info$ {} \; ./x11-toolkits/xview-clients/pkg-plist ./x11-toolkits/xview/pkg-plist ./x11-toolkits/gnustep-gui/pkg-plist ./lang/mlton/pkg-plist ./lang/rexx-imc/pkg-plist ./japanese/texinfo/pkg-plist ./japanese/jishyo/pkg-plist ./misc/xfce4-weather-plugin/pkg-plist ./misc/findutils/pkg-plist ./misc/gnustep-examples/pkg-plist ./graphics/pixen/pkg-plist ./graphics/qgis/pkg-plist ./graphics/laternamagica/pkg-plist ./graphics/gmt/pkg-plist ./graphics/zgv/pkg-plist ./graphics/gnustep-slideshow/pkg-plist ./graphics/graphos/pkg-plist ./graphics/preview/pkg-plist ./security/libgpg-error/pkg-plist ./security/libtasn1/pkg-plist ./x11-fm/krusader2/pkg-plist ./science/paje/pkg-plist ./science/fisicalab/pkg-plist ./science/gtamsanalyzer/pkg-plist ./x11-clocks/aclock/pkg-plist ./ftp/gnustep-ftp/pkg-plist ./math/edenmath/pkg-plist ./math/asymptote/pkg-plist ./math/scilab/pkg-plist ./math/wingz3/pkg-plist ./games/nextgo/pkg-plist ./games/gshisen/pkg-plist ./games/gnustep-ladder/pkg-plist ./games/gnuchess/pkg-plist ./games/bsdgames/pkg-plist ./games/fortunate/pkg-plist ./games/rtb/pkg-plist ./games/daimonin/pkg-plist ./emulators/vice/pkg-plist ./emulators/uae/pkg-plist ./java/linux-sun-jdk17/pkg-plist ./java/netbeans-devel/pkg-plist ./java/netbeans/pkg-plist ./java/linux-oracle-jdk18/pkg-plist ./mail/mew/pkg-plist ./mail/addresses/pkg-plist ./databases/datamodeler/pkg-plist ./databases/sqldeveloper/pkg-plist ./databases/grass/pkg-plist ./net-im/freetalk/pkg-plist ./multimedia/pyjama/pkg-plist ./multimedia/poe/pkg-plist ./multimedia/openh264/pkg-plist ./x11-wm/olvwm/pkg-plist ./x11-wm/amiwm/pkg-plist ./print/gspdf/pkg-plist ./print/texlive-docs/pkg-plist ./x11/innerspace/pkg-plist ./net/remotedesk/pkg-plist ./www/grr/pkg-plist ./www/linux-firefox/pkg-plist ./www/drush/pkg-plist ./www/drupal6/pkg-plist ./www/drupal7/pkg-plist ./www/gallery3/pkg-plist ./www/qupzilla-qt4/pkg-plist ./www/qupzilla-qt5/pkg-plist ./archivers/zipper/pkg-plist ./astro/foxtrotgps/pkg-plist ./devel/openocd/pkg-plist ./devel/gorm/pkg-plist ./devel/flyspray/pkg-plist ./devel/projectcenter/pkg-plist ./devel/mkcmd/pkg-plist ./devel/gengetopt/pkg-plist ./devel/cherivis-devel/pkg-plist ./devel/boost-docs/pkg-plist ./devel/liblouis/pkg-plist ./devel/mpatrol/pkg-plist ./sysutils/usermin/pkg-plist ./sysutils/webmin/pkg-plist ./sysutils/batmon/pkg-plist ./sysutils/timemon/pkg-plist ./sysutils/virtualmin/pkg-plist ./sysutils/logstash/pkg-plist ./audio/gnustep-cdplayer/pkg-plist ./audio/mpdcon/pkg-plist ./deskutils/gnustep-notebook/pkg-plist ./deskutils/gworkspace/pkg-plist ./deskutils/gworkspace-gwmetadata/pkg-plist ./deskutils/displaycalibrator/pkg-plist ./deskutils/hamster-applet/pkg-plist ./deskutils/simpleagenda/pkg-plist ./deskutils/cartotheque/pkg-plist ./deskutils/systempreferences/pkg-plist ./textproc/easydiff/pkg-plist ./textproc/popup/pkg-plist ./textproc/scss/pkg-plist ./textproc/linux-f10-aspell/pkg-plist ./textproc/linux-c6-aspell/pkg-plist ./news/fidogate/pkg-plist ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Poudriere build fails on several ports with leftovers message for info/dir
This seems to be the serial error related to turned off docs. /etc/make.conf has these unsets: OPTIONS_UNSET=DOCS HTMLDOCS HELP PORTDOCS APIDOCS RDOC MANPAGES EXAMPLES PORTEXAMPLES Here is the message: Error: Files or directories left over: info/dir Example ports: security/libassuan security/libgcrypt security/libksba poudriere is run with -t of course Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Pourdriere produces faulty build results due to bsd.openssl.mk bug
The is also this bug in bsd.openssl.mk that lines like this CFLAGS+=-Wl,-rpath,${OPENSSLRPATH} need to have -L${OPENSSLRPATH} before -Wl,-rpath. This is because pure -Wl,-rpath,${SOMEPATH} w/out preceding -L${SOMAPATH} doesn't force-link to that path. -L/usr/local/lib is *usually* but *not always* present. I saw the cases when -Wl,-rpath fails to link to that path w/out -L before it. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Replacing vlc with vlc-qt4 broke pkg upgrade
On 03/31/2015 21:29, Kevin Oberman wrote: I don't see any information that vlc-qt4 has replaced vlc. It is just a version of vlc that will build using the old version of Qt. If you update vlc, it will install the required Qt5 ports. vlc in my case is an automatic dependency of multimedia/phonon-vlc which was recently switched to vlc-qt4. In such case 'pkg upgrade' should automatically delete vlc and install vlc-qt4, but pkg fails instead. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Pourdriere produces faulty build results due to bsd.openssl.mk bug
I implemented the new 'basemix' stage-qa test: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=199106 Currently it complains about libssl.so libcrypto.so from base. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Pourdriere produces faulty build results due to bsd.openssl.mk bug
On 04/01/2015 14:30, Guido Falsi wrote: Some examples of things that could (and will) go wrong: - ports silently linking to base ssl due to weirdness in their build scripts - ports linking to other libraries in base which are linked to base ssl, causing conflicts - ports interacting with base parts, which are using base ssl. I am going to write the stage-qa test checking for just that: no base libraries of certain kinds are mixed into the link list. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Pourdriere produces faulty build results due to bsd.openssl.mk bug
On 04/01/2015 14:17, Jung-uk Kim wrote: I know bsd.openssl.mk has been broken for very long time. For example, http://docs.freebsd.org/cgi/mid.cgi?50108FEF.3030405 However, I am not sure whether entirely removing it is the best way going forward. I mean, removing of the dependency on base. Ports should use only openssl port. If you think this isn't a good idea, and ports should still occasionally use base openssl, would you care to explain why you think so? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Pourdriere produces faulty build results due to bsd.openssl.mk bug
I found that packages produced by poudriere likely link with base openssl, while port make likely links with the port openssl. This is because of the lines in bsd.openssl.mk which check for the presence of openssl shared library and headers under PREFIX, and set WITH_OPENSSL_BASE when they aren't present. In case of port make files are likely present, and in case of poudriere build files are likely not present. Example is ftp/curl (with GSSAPI=NONE, OPENSSL=yes options). *Poudriere produces curl library, that causes VirtualBox to break* due to the openssl base vs. port conflicts. See recent emulation@ ML threads. Port make produces curl library that works fine with VirtualBox. I think both WITH_OPENSSL_BASE and WITH_OPENSSL_PORT should be retired, and code checking file presence also should be removed, and all ports should be made to build with an openssl port instead. Ports should never use base OpenSSL. Only 100 ports touch WITH_OPENSSL_... variables. Somebody who is able to make such decisions and has the commit bit should bit should look into this. Otherwise, massively faulty package repositories are produces. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Pourdriere produces faulty build results due to bsd.openssl.mk bug
On 04/01/2015 15:08, Guido Falsi wrote: This would be good. What I am most worried about, though, are problems showing up at runtime only on specific configurations or use cases. This would help nailing them down. Such stage-qa check will currently likely fail on many ports. It can help in testing after the removal of the said dependencies on base OpenSSL, and should be permanently enabled afterwards. Currently VirtualBox is broken when installed out of the default package repository. I also observed that rebuilding only curl locally doesn't help for other broken things, and one also needs to rebuild kde4-network, and maybe other things. So the problem is quite nasty. It has been triggered by some curl update during March. Check the last entry about asterisk in UPDATING too (I maintain it and that one did hit me on a server of mine) Yes, things shouldn't need to be disabled just because of this conflict. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Replacing vlc with vlc-qt4 broke pkg upgrade
Few days ago vlc was replaced with vlc-qt4. I rebuilt all packages from my ports tree with poudriere, tried to 'pkg upgrade', and pkg fails to handle this: All repositories are up-to-date. Checking for upgrades (28 candidates): 100% Processing candidates (28 candidates): 100% Checking integrity... done (1 conflicting) pkg: Cannot solve problem using SAT solver: conflict rule: The following packages conflict with each other: vlc-2.2.0_3,4(r), vlc-qt4-2.2.0_3,4(r) dependency rule: package vlc(r) depends on: vlc(l)phonon-vlc(l) conflict rule: The following packages conflict with each other: vlc-2.2.0_3,4(r), vlc-2.2.0_3,4(r) upgrade rule: upgrade local vlc-2.2.0_2,4 to remote vlc-2.2.0_3,4 cannot install package vlc, remove it from request? [Y/n]: vlc is installed as automatic. Shouldn't pkg uderstand that vlc and vlc-qt are counterparts and should be substituted? There is no /usr/ports/UPDATING record for this. Is think such record should be added if pkg can't handle this. Or what is the solution? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Why pkg-install is run during port build?
pkg-install is run once during port's 'make', and another time during port's 'make install'. Is this intended? pkg-install is for the package, and port make didn't build/install the package. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Why pkg-install is run during port build?
On 03/30/2015 14:42, Matthew Donovan wrote: With FreeBSD packages and ports are the same. As it is tje same database No, package and port are different things. Ports are the recipes to build packages or install directly, they are in file system under /usr/ports, and packages are the collection of binaries that can be installed, located in the repository, like one that is being pointed to by /etc/pkg/FreeBSD.conf My question is why pkg-install is run during the stage phase? Isn't this the wrong thing to do? I think only make install should run it. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Why pkg-install is run during port build?
On 03/30/2015 16:16, Baptiste Daroussin wrote: And pkg-install is not run at all during stage if some ports are explicitly doing that they should be fixed! That was it, thanks, it was called explicitly. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Firefox 36.0.4 seg fault
On 03/24/2015 12:45, Todd Michel McComb wrote: Firefox built cleanly, segmentation fault when trying to run. No other info returned, so very hard to debug. (I have built Firefox before, but it has been a few versions since I updated it.) What does this command print? $ xdpyinfo | grep GLX Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
[SUGGESTED FEATURE] USE_SVNREPO allows to build the port using only subversion repository URL
I came across one package which doesn't distribute source tarballs, and only offers subversion URL. I implemented the new feature USE_SVNREPO that allows to build the port in such setup. Here is what it does: * exports source from the subversion repository for specified URL/revision * resets the timestamp of every file and directory to 1970-01-01 for reproducibility * creates tarball Patch adding this feature is attached to this bug report: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=198449 Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Single worst bug for ports: lang/gcc* doesn't support c++11
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=193528 Many packages still don't build with clang (in part because Linux didn't adopt it as much as FreeBSD did), and at the same time they are changing code to be c++11. As a result, many such packages can't be updated any more. In two days I hit this problem twice. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Single worst bug for ports: lang/gcc* doesn't support c++11
On 03/07/2015 21:28, Russell L. Carter wrote: I would like to understand better the problem here, because I use c++11 features heavily with lang/gcc49 on a daily basis with zero problems. No, gcc-4.9.3 fails in the same way. Specific missing feature: std::snprintf error: 'snprintf' is not a member of 'std' Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
pkg fails on assertion
Hello I have problem with pkg during packages update #pkg --version 1.4.0.beta2 #pkg upgrade Updating pcbsd-major repository catalogue... pcbsd-major repository is up-to-date. Updating pcbsd-sysrel repository catalogue... pcbsd-sysrel repository is up-to-date. All repositories are up-to-date. Checking for upgrades (1355 candidates)... done Processing candidates (1355 candidates)... done Checking integrity...Assertion failed: (pkgdb_ensure_loaded(j-db, p2, PKG_LOAD_FILES|PKG_LOAD_DIRS) == EPKG_OK), function pkg_conflicts_need_conflict, file pkg_jobs_conflicts.c, line 211. Child process pid=66300 terminated abnormally: Abort trap OS is FreeBSD (PC-BSD) 10.1 Packages was downloaded previously (about 1Gb, many packages). If you need to see something else please let me know. -- Best regards, Yuri Momotyuk ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Update of security/gnupg fails because of conflict with security/dirmngr
=== Registering installation for gnupg-2.1.0_1 as automatic pkg-static: gnupg-2.1.0_1 conflicts with dirmngr-1.1.0_12 (installs files into the same place). Problematic file: /usr/local/bin/dirmngr UPDATING has no info about this. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Update of security/gnupg fails because of conflict with security/dirmngr
On 11/29/2014 04:56, A.J. 'Fonz' van Werven wrote: Yes, that sort of thing seems to be happening a lot lately. I found out that it often helps if you temporarily uninstall the offending port, then install the other one and finally reinstall the uninstalled one. Or in your case: 1. delete dirmngr (remember if it takes anything else with it!) 2. install gnupg; 3. reinstall dirmngr again (as well as anything that got deleted with it). Hope that helps. However, considering that this sort of thing occurs commonly these days, I suspect there must be something wrong in the ports infrastructure. What I also find somewhat mindboggling is how ports are built, staged and packaged entirely, only THEN to discover there's a conflict. Couldn't that have been detected way earlier? AvW I think, what typically happens is that one package depends on another, and that dependency package is just imported in the dependent one, while keeping the same file names installed. In this case it looks like gnupg requires dirmngr, and parts of it were imported into gnupg, while dirmngr keeps living its own life in the system. This is really an evil practice, but people do this, usually without causing conflicts. Chrome has a whole lot of packages included, and also mplayer has. In this case gnupg port should rename offending files, nothing else would solve the problem. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Update of security/gnupg fails because of conflict with security/dirmngr
Here is the workaround: sqlite3 /var/db/pkg/local.sqlite delete from deps where name = 'dirmngr' pkg delete dirmng Then gnupg would update successfully. security/dirmngr moved into gnupg, and should be labeled as deleted. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
'portupgrade' is broken by the recent ports tree change
I just updated the ports tree, and portupgrade now breaks. There is nothing about this in UPDATING. # portupgrade -a /usr/local/lib/ruby/site_ruby/1.9/pkgtools/portsdb.rb:118:in `block (2 levels) in fill': MOVED file format error (PortsDB::MOVEDError) from /usr/local/lib/ruby/site_ruby/1.9/pkgtools/portsdb.rb:112:in `each' from /usr/local/lib/ruby/site_ruby/1.9/pkgtools/portsdb.rb:112:in `block in fill' from /usr/local/lib/ruby/site_ruby/1.9/pkgtools/portsdb.rb:111:in `open' from /usr/local/lib/ruby/site_ruby/1.9/pkgtools/portsdb.rb:111:in `fill' from /usr/local/lib/ruby/site_ruby/1.9/pkgtools/portsdb.rb:106:in `initialize' from /usr/local/lib/ruby/site_ruby/1.9/pkgtools/portsdb.rb:181:in `new' from /usr/local/lib/ruby/site_ruby/1.9/pkgtools/portsdb.rb:181:in `setup' from /usr/local/lib/ruby/site_ruby/1.9/pkgtools/pkgtools.rb:242:in `init_pkgtools_global' from /usr/local/sbin/portupgrade:530:in `block in main' from /usr/local/lib/ruby/1.9/optparse.rb:882:in `initialize' from /usr/local/sbin/portupgrade:237:in `new' from /usr/local/sbin/portupgrade:237:in `main' from /usr/local/sbin/portupgrade:2371:in `main' portupgrade-2.4.12_2,2 Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Why USE_PYTHON=yes doesn't rebuild default python meta-port?
I am rebuilding the port hierarchy with the top port having USE_PYTHON=yes. This package runs python-config. But only python2 and python27 were actually built, and therefore only python2-config is available and my port breaks. Default meta-port python-2.7, that would have contained python-config, wasn't built. But USE_PYTHON=yes asks for the default python, doesn't it? So is this a bug in the make scripts, or else why default python version wasn't built? FreeBSD-10 with current ports tree. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Unable to upgrade to libdmtx-0.7.4_5
On 06/24/2014 03:18, Dr. Peter Voigt wrote: After yesterday ports tree update I am unable to upgrade libdmtx: # portmaster --no-confirm --no-term-title -D -G libdraw There is a problem in this port's dependency on ImageMagick line. I notified maintainer. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
kde4 package isn't available on pkg.freebsd.org
No 'latest' version for 10 and 9 (ex. http://pkg.freebsd.org/freebsd:10:x86:64/latest/All/) Is there a way to see the logs to determine what exactly is holding it up? There is this mailing list http://lists.freebsd.org/pipermail/freebsd-pkg-fallout/ but it is a plain list of failures for all versions and times. Lacking kde4 package (and also gnome2 and other popular packages) is a problem because it likely prevents people from upgrading their systems. And local rebuild of kde may take a long time. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
multimedia/xine maintainer should be multimedia@, not ports@
Maybe somebody can fix this real quick? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Conflicts between libcrypto.so.8 from port vs. libcrypto.so.6 from the base
While tracking down the crash in libssl on 9.2, I found that it was due to this conflict. First problem is that curl that I build always links with libcrypto.so.8 installed by openssl-1.0.1_9, and curl coming from the repository comes built with libcrypto.so.6 from the base system. There is also the warning during curl build: /usr/bin/ld: warning: libcrypto.so.6, needed by /usr/lib/libheimntlm.so, may conflict with libcrypto.so.8 openssl is installed because a bunch of other ports depend on it. So one problem is this ftp/curl build volatility (it links to libcrypto.so.8, when security/openssl is installed even though it doesn't depend on it). The second problem is that the program that depends on two ports, one of which is built with security/openssl, and another one depends on the base openssl, will always crash due to this conflict. And some packages are built with USE_OPENSSL and some aren't. Shouldn't every port that needs libcrypto/libssl always depend on security/openssl to avoid such problems? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Package builder keeps sending fallout failure messages for the problems that have been long fixed
As a maintainer of perftools, I am getting the messages like this: [REL - 10i386-quarterly][devel/google-perftools] Failed for google-perftools-1.8.3 in build ... Ident: $FreeBSD: branches/2014Q1/devel/google-perftools/Makefile 327723 2013-09-20 17:04:43Z bapt $ Log URL: http://beefy1.isc.freebsd.org/bulk/10i386-quarterly/2014-02-12_01h01m52s/logs/google-perftools-1.8.3.log Build URL: http://beefy1.isc.freebsd.org/bulk/10i386-quarterly/2014-02-12_01h01m52s ... Makefile ident: $FreeBSD: branches/2014Q1/devel/google-perftools/Makefile 327723 2013-09-20 17:04:43Z bapt $ The version 1.8.3 had problems at the time, but this has been fixed, and now it is a new version 2.X which works. I am not sure, am I supposed to do something with this? Why does the system now, in Feb 2014, still build packages with sources from Sep 2013? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: What is the problem with ports PR reaction delays?
On 01/25/2014 11:38, Baptiste Daroussin wrote: We just recovered from cvs-svn switch I think noone at all is willing to do the job again. Git has lots of drawbacks and if badly handled with all the above, it will be a real nightmare. and That is said from someone who likes git (except that the UI can get overcomplicated) Git has some features that aren't in svn, and also svn has some features that aren't in fit, though to the lesser extent. There is no reason why enough important missing features can't be added to both systems over time so that seamless back and forth data migration would be possible. Because they basically do the same thing. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: libstdc++.so.6 in base 9.2 system collides with libstdc++.so.6 in gcc47
On 01/20/2014 03:13, John Marino wrote: The approach taken by lang/gcc-aux mirrors your sentiment in the PRs. The rpath is inserted by the compiler itself, automatically, for every program and shared object produced. The individual ports would not need to set rpath themselves if they were built with lang/gcc-aux rather than lang/gcc47. This wouldn't work for bound to C/C++ projects in interpreted languages though. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: What is the problem with ports PR reaction delays?
On 01/24/2014 20:16, Alfred Perlstein wrote: (maybe there is some great ports system that I'm not aware of that makes this all as easy github, but I somehow doubt that.) github itself is closed source, but 95% of its functionality is based on git which is open. One only needs to invoke 3-4 git operations to support what it does on the website side. Register on the site, fork the project under user's login, submit a pull request, merge a fork's branch to the main branch. All these are basically git commands. Without the glossiness of github, this is not that large of a project. Submitters will do the rest through git. I think, instead of tediously going through the PRs by hand, it is wiser to set up some system like this. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: What is the problem with ports PR reaction delays?
On 01/25/2014 05:43, Bernhard Fröhlich wrote: With the scripts it should be possible to fetch the patch of a PR, apply and commit it to your redports repository and do additional changes until you are okay with it. What is still missing is a script that helps committing the changes to the FreeBSD tree but it's really not that complicated to write that. This sounds like a very manual process again. Today it is already manual. This only automates the patching step. be more specific and name the problem, if you mentioned it? The raised concerns were that automatic build testing might result in less testing on committer side. I agree that this might happen and automatic build testing is only one part of what committers need to do. A huge backlog and no response for months is definitely nothing we want. But what exactly do committers test that can't be automated? Automatic system could install the port with dependencies into the blank installation, check syntax correctness, run 'lint' on it, verify that it installs/uninstalls cleanly and doesn't leave residue, etc etc. What is beyond that? I am trying to understand, what would the general ports expert without the intimate knowledge of this particular package catch, that automated system won't be able to catch? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: What is the problem with ports PR reaction delays?
On 01/25/2014 14:44, Aryeh Friedman wrote: The key seems to be that no one has time to do the stuff they really want to do (get new ports into the system)... to that end automating everything that can be automated is sure help free up comitter time so they can look at what is interesting Yes. I just can't imagine any generic port tests that can't be automated and coded into the script once and for good. Ideal system should be like github with the added automated testing between pull request submission and merge. It should either fail and notify the submitter, or succeed and notify the committers. Today all committers do for ports is running some generic tests by hand, like 'lint' with various flags. Install/uninstall/etc. No wonder this isn't very a rewarding activity, and committers probably perceive it as a necessary evil. The way how it is, it causes a waste of their time. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: What is the problem with ports PR reaction delays?
On 01/25/2014 15:48, Aryeh Friedman wrote: Git hup (or*ANY* remote service for that matter) is a no go IMO But both Debian and Fedora do this with automated remote testing, and they don't seem to complain. How is our ports different in this respect? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: USE_GCC doesn't set rpath
On 01/21/2014 15:31, Shane Ambler wrote: I think you will find that qbittorrent will need to be built with the same gcc version as libtorrent-rasterbar. I believe qbittorrent is loading /usr/lib/libstdc++.so.6 then it loads libtorrent-rasterbar.so.7 which tries to load libstdc++ and it is given the already open copy which doesn't have GLIBCXX_3.4.15 that it needs to run. Yes, you are right, thst's what is happening. So if any dependency has USE_GCC set, all dependent packages should also have USE_GCC set. Can ports build infrastructure automatically set USE_GCC for dependent ports? Because doing this by hand in each port is tedious and error prone. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: USE_GCC doesn't set rpath
On 01/23/2014 11:56, John Marino wrote: While I basically agree with this sentiment, is not it just ports that use C++ that are affected? Could not this be narrowed down to We need a single compiler for ports that use C++ ? It's not necessarily just C++. Python ports bound to C++ libraries with USE_GCC would create even more grave problem. It will be needed to create a special version of python with this rpath in it. And if there is a mix of different versions of gcc in various dependencies it would be just impossible to run such program due to conflicts. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
libstdc++.so.6 in base 9.2 system collides with libstdc++.so.6 in gcc47
After the ports update my qbittorrent fails due to this collision: $ qbittorrent /usr/lib/libstdc++.so.6: version GLIBCXX_3.4.15 required by /usr/local/lib/libtorrent-rasterbar.so.7 not found I guess, all packages depending on gcc47 libs would likely fail the same way. gcc probably forgot to bump so revision. The workaround is to run with LD_LIBRARY_PATH=/usr/local/lib/gcc47 But what is the general solution for this? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: libstdc++.so.6 in base 9.2 system collides with libstdc++.so.6 in gcc47
On 01/19/2014 14:07, John Marino wrote: The gcc47 should be built with specs that automatically insert a runpath (-Wl,-rpath,) of /usr/local/gcc47/ so that every binary and library built by it knows where to look. This is what lang/gcc-aux does. Looks like it doesn't do this. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Please check in the new port for Bitmessage
This is a new near-instant messenger replacing traditional mail and IM clients. It allows to deeply encrypt the messages, therefore avoiding message snooping. http://www.freebsd.org/cgi/query-pr.cgi?pr=184322 http://en.wikipedia.org/wiki/Bitmessage Thank you, Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
'pkg upgrade' produces 'conflict found' warnings
Proceed with upgrading packages [y/N]: y Checking integrity... Conflict found on path /usr/local/bin/a2p between perl-5.16.3_1(lang/perl5.16) and perl-5.14.4_2(lang/perl5.14) Conflict found on path /usr/local/bin/c2ph between perl-5.16.3_1(lang/perl5.16) and perl-5.14.4_2(lang/perl5.14) Conflict found on path /usr/local/bin/config_data between perl-5.16.3_1(lang/perl5.16) and perl-5.14.4_2(lang/perl5.14) Conflict found on path /usr/local/bin/corelist between perl-5.16.3_1(lang/perl5.16) and perl-5.14.4_2(lang/perl5.14) ... Binary upgrade fails this way. This system is 9.2-STABLE with PKGNG in use, and perl has been upgraded to perl5-5.16.3_1. So I assume conflicts come from the remote packages? This command has been run, if this matters: pkg set -o lang/perl5.14:lang/perl5.16 This message should be clarified: where the conflict is found exactly? The current message isn't user friendly. User didn't do anything wrong, and the message doesn't add clarity as to what the problem is. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Should /usr/bin/perl be a link to /usr/local/bin/perl ?
I found that many ports specify /usr/bin/perl as an interpreter. This comes from Linux. Examples: valgrind-snapshot, windowmaker, enscript-a4, a2ps, svgalib /usr/bin/perl isn't installed by perl port. There are several solutions, in the order of increasing complexity of solution: 1. Install the link /usr/bin/perl (hackish, but it will fix many broken ports in one shot) 2. Make a package scripts check for interpreter and break the install of offending packages 3. Fix all offending packages Which solution should be preferred in your opinion? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Should /usr/bin/perl be a link to /usr/local/bin/perl ?
On 10/21/2013 09:53, andrew clarke wrote: Are you aware the Perl interpreter ports already have a make option to create symlinks in /usr/bin? Hm, I wasn't aware of this. And I do have this option set in port, never touched it, and still don't have /usr/bin/perl. Mistery. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: pkgng+portmaster: portmaster erased the field packages.time
On 08/22/2013 04:49, Bryan Drewery wrote: By Packages.time I assume you mean the timestamp the package was installed? There is no conversion timestamp stored by pkgng. After pkg2ng has ran, all fields packages.time were !=null. This was probably the original install time then. How are you seeing this timestamp? With 'pkg query %o %t'? I use firefox plugin to look into .sqlite file directly. Portmaster itself is not modifying this data at all (it can't), only pkgng updates this timestamp. After portmaster ran, I observed all package.time fields became null, except for the ones updated by portmaster. I am not sure who deleted them, this is what happened after portmaster ran. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: pkgng+portmaster: portmaster erased the field packages.time
On 08/22/2013 04:49, Bryan Drewery wrote: By Packages.time I assume you mean the timestamp the package was installed? There is no conversion timestamp stored by pkgng. After pkg2ng has ran, all fields packages.time were !=null. This was probably the original install time then. How are you seeing this timestamp? With 'pkg query %o %t'? I use firefox plugin to look into .sqlite file directly. Portmaster itself is not modifying this data at all (it can't), only pkgng updates this timestamp. After portmaster ran, I observed all package.time fields became null, except for the ones updated by portmaster. I am not sure who deleted them, this is what happened after portmaster ran. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
pkgng+portmaster: portmaster erased the field packages.time
After I converted to pkgng, field packages.time contained the time of conversion. When I ran portmaster to rebuild few ports, all those timestamps disappeared, and portmaster only left the timestamps on the ports it reinstalled. I suspect this is a bug, since this behavior doesn't appear reasonable. Why would portmaster erase the time field? How can it conclude it is wrong? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Why the port build log file has terminal escape sequences?
On 02/03/2013 03:13, Chris Rees wrote: I guess this is a cmake/clang port? Cmake, clang and several other tools give colorised output. This is achieved with escape sequences. I understand about colorized output. But aren't such tools normally expected to turn colors off when run not in terminal? This is achieved by checking isatty (ttyname(3)). Probably I should file such PR to cmake. This might be a regression there. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Why the port build log file has terminal escape sequences?
I redirect the log into the file: portupgrade -a 21 | tee portupgrade.log However, log still has escape sequences as if it was printed to the terminal: ^[[0m^[[34m^[[1mGenerating moc_mainwindow.cpp^M ^[[0m^[[34m^[[1mGenerating subscriptionlistdelegate.moc^M ^[[0m^[[34m^[[1mGenerating knconfigwidgets.moc^M ^[[0m^[[34m^[[1mGenerating previewdialog.moc^M ^[[34m^[[1mGenerating entriescountdialog.moc^M I believe this started not long ago. What is wrong? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Please check in the PR#173730: [PATCH] Added rc.d script for net/vnstat port
http://www.freebsd.org/cgi/query-pr.cgi?pr=173730 I contacted the maintainer Dennis Herrmann dhn@ on Nov 09, 2012 and got no response in 2+ months. This PR is also assigned to dhn@ with no response. What is the policy of detecting unresponsive maintainers? There are quite a few ports owned by dhn @. Should all of them be reassigned to ports @ ? Yuri ---list of ports owned by dhn@--- astro/libnova audio/mp3burn audio/cripple audio/last-cmus audio/solfege audio/p5-Audio-Mixer audio/cmus-post.fm audio/ncmpcpp audio/p5-Audio audio/vitunes audio/xpi-musicpm audio/cutmp3 audio/snd comms/picocom comms/ncid comms/viewfax devel/p5-Data-Table devel/p5-Date-Simple devel/libdsp devel/synfig devel/rubygem-trollop devel/rubygem-lockfile devel/p5-Devel-Events devel/trio devel/skalibs devel/p5-Module-Loaded devel/etl devel/nasm devel/p5-Data-TreeDumper devel/hypersrc editors/poedit ftp/ftpsync ftp/scythia games/rubix games/doom-freedoom games/zoom games/vitetris graphics/gocr graphics/fotofix graphics/synfigstudio graphics/silgraphite irc/anope irc/py-supybot irc/simpleirc irc/py-supybot-plugins irc/miau lang/mmix mail/elmo mail/elmo-devel mail/p5-Mail-Ezmlm math/fxt misc/hello multimedia/youtube-viewer multimedia/smplayer multimedia/smplayer-themes net/p5-Net-Libdnet net/p5-Net-ARP net/libosip net/wired net/radvd net/vnstat net/rubygem-rubytter net/p5-XML-Fast net/wired-tracker net/wire net/GeoIP net-im/tkabbur net-im/rubygem-termtter net-im/mcabber net-mgmt/rrdbot net-mgmt/cowpatty science/clhep sysutils/fusefs-curlftpfs sysutils/fusefs-gstfs sysutils/ttyload sysutils/rubygem-teamocil sysutils/logtool textproc/rss2html textproc/rubygem-termcolor textproc/p5-Tk-XMLViewer textproc/smu textproc/p5-XML-Parser-Lite-Tree www/zerowait-httpd www/p5-HTML-Template-Associate www/p5-WWW-FreeProxy www/multisort www/ruby-nora www/p5-WWW-TinySong x11/i3lock x11/lsw x11-fm/gentoo x11-fonts/montecarlo_fonts x11-toolkits/p5-Tk-FontDialog x11-wm/ratpoison x11-wm/wmfs x11-wm/musca x11-wm/i3 x11-wm/antiwm ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Why submitted PRs don't show up on the web interface right away?
Yesterday I submitted the PR and got an acknowledgement e-mail with this link: http://www.freebsd.org/cgi/query-pr.cgi?pr=173710 Usually in such situation this link shows No PRs Matched Query in this URL for a few hours and then info shows up. But not in this case. Now, after 18+ hours, this link still shows No PRs Matched Query. Few questions: * What changed? Why PR doesn't show in the web interface? * Why don't PRs show up online immediately? Is there some process that synchronizes something in batches? What is the point of this process? * Why FreeBSD submission process doesn't require authentication? * I think the captcha in use is very outdated and will be easily broken once some hackers will come across it. I think google recaptcha would be a very welcome change. Or maybe something a bit more creative like https://github.com/josscrowcroft/MotionCAPTCHA would also be good. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Let's talk about subversion/svn
On 11/19/2012 10:40, Paul Schmehl wrote: It's not just subversion. I was working on a new port the other day, and one of it's dependencies triggered a lengthy traverse of numerous x11, x11-tookits, x11-fonts, etc., etc. Rather mind-boggling when you realize this is a program that generates graphs rather than a GUI-based program. It would be very beneficial if somebody made a program that would visualize the particular port's dependencies as a graph (like .dot format). This should be quite easy to do. Or maybe such program already exists? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Something wrong with mirror.irssi.org? (Was: Re: svn commit: r302888 - head/irc/irssi-devel)
On Tue, 21 Aug 2012 18:48:59 + (UTC), Beat Gaetzi wrote: Author: beat Date: Tue Aug 21 18:48:59 2012 New Revision: 302888 URL: http://svn.freebsd.org/changeset/ports/302888 Log: - Mark BROKEN: size mismatch fetch: http://mirror.irssi.org/snapshots/irssi-20120624.tar.gz: size mismatch: expected 1331265, actual 1331625 Reported by: pointyhat Modified: head/irc/irssi-devel/Makefile Modified: head/irc/irssi-devel/Makefile == --- head/irc/irssi-devel/Makefile Tue Aug 21 18:46:31 2012 (r302887) +++ head/irc/irssi-devel/Makefile Tue Aug 21 18:48:59 2012 (r302888) @@ -16,6 +16,8 @@ COMMENT= A modular IRC client with many LIB_DEPENDS= glib-2.0:${PORTSDIR}/devel/glib20 +BROKEN=size mismatch + CONFLICTS=irssi-0.8* GNU_CONFIGURE=yes I've filed this as 170541, but it looks like filesize changes every now and then (something wrong with mirror.irssi.org?). ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: FreeBSD ports which are currently scheduled for deletion
On Wed, 08 Feb 2012 02:12:39 -0800, perryh wrote: lini...@freebsd.org wrote: portname: graphics/vrml2pov description:Convert VRML files to POVRay source maintainer: po...@freebsd.org status: BROKEN deprecated because: unfetchable This seems to be a ports-infrastructure problem, rather than a problem in the vrml2pov port. make fetch-recursive reports fetch: http://www.chemicalgraphics.com/paul/vrml2pov/dist/v0.7/vrml2pov-src.zip: Moved Temporarily and gives up after the attempt from ftp.FreeBSD.org also fails. However, Firefox successfully downloads the distfile from the chemicalgraphics.com URL. The infrastructure needs to handle redirections. Infrastructure actually explicitly disallows redirects with 'A' flag in FETCH_ARGS. If you can't find the URL where it redirects to, you could specify FETCH_ARGS=-Fpr for such ports. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: [HEADSUP][CFT] pkgng beta1 is out
On Wed, 8 Feb 2012 15:07:16 +, Anton Shterenlikht wrote: On Wed, Feb 08, 2012 at 01:21:12PM +0100, Baptiste Daroussin wrote: cc1: warnings being treated as errors usergroup.c: In function 'pkg_add_user_group': usergroup.c:75: warning: implicit declaration of function 'gr_init' usergroup.c:75: warning: nested extern declaration of 'gr_init' usergroup.c:76: warning: implicit declaration of function 'gr_lock' usergroup.c:76: warning: nested extern declaration of 'gr_lock' usergroup.c:77: warning: implicit declaration of function 'gr_fini' usergroup.c:77: warning: nested extern declaration of 'gr_fini' usergroup.c:80: warning: implicit declaration of function 'gr_tmp' usergroup.c:80: warning: nested extern declaration of 'gr_tmp' usergroup.c:84: warning: implicit declaration of function 'gr_copy' usergroup.c:84: warning: nested extern declaration of 'gr_copy' usergroup.c:88: warning: implicit declaration of function 'gr_mkdb' usergroup.c:88: warning: nested extern declaration of 'gr_mkdb' *** Error code 1 Your current is too old, could you please try to update it? the gr_* function where committed in r228545 version. Thanks, I got it working on r231193. I applied the portmaster patch from git, then running portmaster -a, I get: === Port directory: /usr/ports/devel/xorg-macros === Starting check for build dependencies === Gathering dependency list for devel/xorg-macros from ports === Launching child to install ports-mgmt/pkg xorg-macros-1.15.0 ports-mgmt/pkg which ends in === Installing for pkg-1.0.b1 ===Generating temporary packing list === Checking if ports-mgmt/pkg already installed ===pkg-1.0.b1 is already installed You may wish to ``make deinstall'' and install this port again by ``make reinstall'' to upgrade it properly. If you really wish to overwrite the old port of ports-mgmt/pkg without deleting it first, set the variable FORCE_PKG_REGISTER in your environment or the make install command line. *** Error code 1 Stop in /usr/ports/ports-mgmt/pkg. What am I doing wrong? You are likely missing use_pkgng=yes in your portmaster.rc. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: [HEADSUP][CFT] pkgng beta1 is out
On Mon, Jan 30, 2012 at 01:39:30PM +0100, Baptiste Daroussin wrote: [...] to use pkgng: echo WITH_PKGNG=yes /etc/make.conf make -C /usr/ports/ports-mgmt/pkg install clean Some links: http://wiki.FreeBSD.org/pkgng http://github.com/pkgng/pkgng Note that on github you can find a patch for portmaster (against 3.10) The patch seems to have typos in it (usr_pkgng): https://github.com/pkgng/pkgng/blob/master/ports/portmaster.patch#L514 https://github.com/pkgng/pkgng/blob/master/ports/portmaster.patch#L528 And most stupid question for today - how do I actually tell portmaster to use pkgng after applying the patch? Thanks, Yuri pgpWRBGFuloAU.pgp Description: PGP signature
libthai-0.1.5: build is broken by failing self tests
Here is the failure log: PASS: test-thcoll.sh Testing with string: ���ʴդ�Ѻ �繡�÷��ͺͧ Total 8 cut points.Cut points list: 6, 10, 14, 18, 21, 23, 26, 29 Output string is ���ʴ�WBR��ѺWBR���WBRWBR���WBR��WBR�ͺWBR���WBR�ͧbump. Output string length is 71 *** End of thbrk self test ** PASS: test_thbrk Testing thwchar... Input: tis-620-0 string of length 32: ���ʴդ�Ѻ �繡�÷��ͺͧ Output: Unicode string of length 32: Convert back to tis-620-0 string... Output: tis-620-0 string of length 32: ���ʴդ�Ѻ �繡�÷��ͺͧ Input = output, correct! Test thwchar OK. PASS: test_thwchar Testing with input string: ���ʴդ�Ѻ �繡�÷��ͺͧ Converting to Unicode... Calling th_wbrk()... Total 8 cut points.Cut points list: 6, 10, 14, 18, 21, 23, 26, 29 Calling th_wbrk_line() Return value from th_wbrk_line is 71 Output string length is 71 Compare with result from th_brk_line().. Error! Comparison of results from th_brk_line() and th_wbrk_line() failed. th_brk_line :���ʴ�WBR��ѺWBR���WBRWBR���WBR��WBR�ͺWBR���WBR�ͧ8.0 lib version bump. th_wbrk_line:���ʴ�WBR��ѺWBR���WBRWBR���WBR��WBR�ͺWBR���WBR�ͧ Test th_wbrk_line() failed... FAIL: test_thwbrk === 1 of 9 tests failed === gmake[2]: *** [check-TESTS] Error 1 gmake[2]: Leaving directory `/usr/ports/devel/libthai/work/libthai-0.1.5/tests' gmake[1]: *** [check-am] Error 2 gmake[1]: Leaving directory `/usr/ports/devel/libthai/work/libthai-0.1.5/tests' gmake: *** [check-recursive] Error 1 *** Error code 1 ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: isc dhcp does not make
On Thu, Jan 26, 2012 at 09:38:22AM +0900, Randy Bush wrote: FreeBSD dfw0.psg.com 9.0-STABLE FreeBSD 9.0-STABLE #0: Wed Jan 25 21:56:55 UTC 2012 r...@dfw0.psg.com:/usr/obj/usr/src/sys/DFW0 amd64 # portupgrade -N net/isc-dhcp42-server [Gathering depends for net/isc-dhcp42-server . done] --- Installing 'isc-dhcp42-server-4.2.3_2' from a port (net/isc-dhcp42-server) --- Building '/usr/ports/net/isc-dhcp42-server' === Cleaning for isc-dhcp42-server-4.2.3_2 === Vulnerability check disabled, database not found === License ISCL accepted by the user === Found saved configuration for isc-dhcp42-server-4.2.3_2 === Extracting for isc-dhcp42-server-4.2.3_2 = SHA256 Checksum OK for dhcp-4.2.3-P2.tar.gz. === Patching for isc-dhcp42-server-4.2.3_2 === Applying extra patch /usr/ports/net/isc-dhcp42-server/files/extra-patch-bind__Makefile === Applying FreeBSD patches for isc-dhcp42-server-4.2.3_2 === isc-dhcp42-server-4.2.3_2 depends on executable: gmake - found === Configuring for isc-dhcp42-server-4.2.3_2 checking for a BSD-compatible install... /usr/bin/install -c -o root -g wheel checking whether build environment is sane... yes checking for a thread-safe mkdir -p... ./install-sh -c -d checking for gawk... gawk checking whether gmake sets $(MAKE)... yes checking whether to enable maintainer-specific portions of Makefiles... no checking for gcc... cc checking for C compiler default output file name... configure: error: C compiler cannot create executables See `config.log' for more details. === Script configure failed unexpectedly. Please report the problem to z...@freebsd.org [maintainer] and attach the /usr/ports/net/isc-dhcp42-server/work/dhcp-4.2.3-P2/config.log including the output of the failure of your make command. Also, it might be a good idea to provide an overview of all packages installed on your system (e.g. an `ls /var/db/pkg`). *** Error code 1 clang? clang what? checking for gcc... cc, doesn't look like clang. Providing the config.log, as the message suggests, would help here. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: isc dhcp does not make
On Thu, Jan 26, 2012 at 10:04:52AM +0900, Randy Bush wrote: [...] configure:2703: cc -O2 -pipe -march=i686 -fPIC -fno-strict-aliasing ^^^ That's why. Make sure you don't define it in /etc/make.conf. conftest.c:1: error: CPU you selected does not support x86-64 instruction set [...] Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Two Minor Ports Patches
On Sun, Jan 15, 2012 at 10:39:48AM -0500, George Mitchell wrote: After csupping to 9.0-STABLE and updating my ports tree, I encountered two compile errors during portmaster -a: hal: USB_GET_REPORT_ID undefined in probe-hiddev.c. See first attached patch (which probably needs a __FreeBSD_version conditional). sessreg: ttyslot no longer exists. See second attached patch which is clearly wrong, but works as a stopgap. -- George Mitchell The sessreg problem should be fixed by running `make delete-old` in /usr/src. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: New mailing list for ports announcements
On Thu, Dec 29, 2011 at 09:23:05AM +0300, Sergey V. Dyatko wrote: On Thu, 29 Dec 2011 00:59:26 -0500 Thomas Abthorpe tabtho...@freebsd.org wrote: At the request of adamw@ (and others) we have setup a ports-announce@ mailing list to try distinguish the usual traffic on the ports@ list vs the announcements that seem to get lost in there. Useless and harmful list, IMO. People who can 'lost' announcements here likely lose it on another list, but there is a chance that people who read ports@ lose announce/CFT from _only_ freebsd-ports-announce. Of course, you can write to multiple lists.. But lets look at current@ stable@ and performance@ archives... tadam! Benchmark (Phoronix): FreeBSD 9.0-RC2 vs. Oracle Linux 6.1 Server. You seem to miss the moderated part.. Is that ok ? I think 'No' ( Tons of spam in [ 3 of N ] ML that I subscribed ) You can subscribe at http://lists.freebsd.org/mailman/listinfo/freebsd-ports-announce It is intended, but not limited, to be a means of communicating portmgr@ announcements, Calls for Testing, plus other relevant information to be used by our committers and ports maintainer community. It is our hope to keep this relatively low in traffic. It is a moderated list, under the auspices of portmgr@. ^ Please subscribe sit back, and enjoy. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: devel/py-lxml fails to build on 10.0-CURRENT
On Fri, Dec 16, 2011 at 01:17:41PM +0400, Ruslan Mahmatkhanov wrote: Good day. I'm trying to build devel/py-lxml 2.3.2 (libreoffice dep) on 10.0-CURRENT with python27. Build fails like this: building 'lxml.etree' extension creating temp.freebsd-10.0-CURRENT-i386-2.7 creating temp.freebsd-10.0-CURRENT-i386-2.7/src creating temp.freebsd-10.0-CURRENT-i386-2.7/src/lxml cc -DNDEBUG -O2 -pipe -fno-strict-aliasing -O2 -pipe -fno-strict-aliasing -fPIC -I/usr/local/include -I/usr/local/include/libxml2 -I/usr/local/include/python2.7 -c src/lxml/lxml.etree.c -o temp.freebsd-10.0-CURRENT-i386-2.7/src/lxml/lxml.etree.o -w In file included from /usr/local/include/libxslt/xsltlocale.h:19, from /usr/local/include/libxslt/xsltInternals.h:24, from src/lxml/lxml.etree.c:265: /usr/include/xlocale.h:45: error: expected declaration specifiers or '...' before 'locale_t' Can anybody confirm that or it's just me? It builded just fine couple of weeks ago. Full buildlog here: http://people.freebsd.org/~rm/py-lxml.txt Thanks in advance. PS. I tried to rebuild libxslt/libxml2 to not avail. Same here. Probably related: http://lists.freebsd.org/pipermail/svn-src-head/2011-November/031517.html Yuri pgpHD3OanEqPc.pgp Description: PGP signature
Re: linphone-base conflicts with ortp and fails to install, it also fails to build: error: expected specifier-qualifier-list before 'UInt96'
On 11/23/2011 11:45, Alberto Villa wrote: Fix committed: ortp dependency has been replaced by linphone-base. Please, update (and check UPDATING). I updated via cvsup (*default release=cvs tag=.) but I don't see the relevant ports UPDATING record. The only recent record there is 2023 and it is related to databases/redis. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: linphone-base conflicts with ortp and fails to install, it also fails to build: error: expected specifier-qualifier-list before 'UInt96'
On 11/23/2011 15:23, Alberto Villa wrote: On Wednesday 23 November 2011 21:34:55 Yuri wrote: I updated via cvsup (*default release=cvs tag=.) but I don't see the relevant ports UPDATING record. The only recent record there is 2023 and it is related to databases/redis. http://www.freebsd.org/cgi/cvsweb.cgi/ports/UPDATING.diff?r1=1.1171;r2=1.1172;f=h Thanks, I got it! cvsup probably has some delay. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: linphone-base conflicts with ortp and fails to install, it also fails to build: error: expected specifier-qualifier-list before 'UInt96'
On 11/21/2011 01:43, Olivier Smedts wrote: Hello, Please look at ports/UPDATING (20111016). I did that, and wrote about this in my message (see below), but command 'portmaster -o net/linphone-base ortp' fails. When I follow instruction in UPDATING and run this command 'portmaster -o net/linphone-base ortp', I get this messages: In file included from ../include/mediastreamer2/msrtp.h:25, from msrtp.c:20: ../include/mediastreamer2/ice.h:60: error: expected specifier-qualifier-list before 'UInt96' Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
linphone-base conflicts with ortp and fails to install, it also fails to build: error: expected specifier-qualifier-list before 'UInt96'
Hi, When I do portupgrade, I get this message (kdenetwork fails to update due to this): === linphone-base-3.2.1_1,1 conflicts with installed package(s): ^M ortp-0.13.0_1^M ^M They install files into the same place.^M You may want to stop build with Ctrl + C.^M === Vulnerability check disabled, database not found^M === License check disabled, port has not defined LICENSE^M ^M === linphone-base-3.2.1_1,1 conflicts with installed package(s): ^M ortp-0.13.0_1^M ^M They will not build together.^M Please remove them first with pkg_delete(1).^M When I follow instruction in UPDATING and run this command 'portmaster -o net/linphone-base ortp', I get this messages: In file included from ../include/mediastreamer2/msrtp.h:25, from msrtp.c:20: ../include/mediastreamer2/ice.h:60: error: expected specifier-qualifier-list before 'UInt96' msrtp.c: In function 'receiver_process': msrtp.c:333: warning: implicit declaration of function 'rtp_get_payload' gmake[5]: *** [msrtp.lo] Error 1 gmake[5]: Leaving directory `/usr/ports/net/linphone-base/work/linphone-3.2.1/mediastreamer2/src' gmake[4]: *** [all] Error 2 gmake[4]: Leaving directory `/usr/ports/net/linphone-base/work/linphone-3.2.1/mediastreamer2/src' gmake[3]: *** [all-recursive] Error 1 gmake[3]: Leaving directory `/usr/ports/net/linphone-base/work/linphone-3.2.1/mediastreamer2' gmake[2]: *** [all] Error 2 gmake[2]: Leaving directory `/usr/ports/net/linphone-base/work/linphone-3.2.1/mediastreamer2' gmake[1]: *** [all-recursive] Error 1 gmake[1]: Leaving directory `/usr/ports/net/linphone-base/work/linphone-3.2.1' gmake: *** [all] Error 2 *** Error code 1 Stop in /usr/ports/net/linphone-base. === make failed for net/linphone-base === Aborting update I am not sure why linphone-base build fails. But it looks like linphone-base contains some version of ortp. And this creates a problem of the conflict: They install files into the same place. Sylvio, should you consider renaming conflicting files stemming from ortp copy into some other name to remove such conflict? Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: portmaster - updating devel/gconf2 triggers install of net/openldap24-sasl-client, even though it's installed and up to date
On Mon, Jun 06, 2011 at 01:07:29PM +0100, Chris Rees wrote: On 6 Jun 2011 12:41, Anton Shterenlikht me...@bristol.ac.uk wrote: # pkg_version -vXs openlda openldap-sasl-client-2.4.25_2 = up-to-date with port # portmaster gconf === Currently installed version: gconf2-2.32.0_2 === Port directory: /usr/ports/devel/gconf2 === Gathering distinfo list for installed ports === Launching 'make checksum' for devel/gconf2 in background === Gathering dependency list for devel/gconf2 from ports === Launching child to install net/openldap24-sasl-client gconf2-2.32.0_2 net/openldap24-sasl-client === Port directory: /usr/ports/net/openldap24-sasl-client === Launching 'make checksum' for net/openldap24-sasl-client in background ^C If I let it continue, I eventually get an error that openldap is already installed. Please advise Did you set any OPTIONS on openldap when you first installed it? I bet the SASL option is set and net/openldap24-client (origin) is installed. We should just kill the openldap2{3,4}-sasl-{client,server} as they introduce all kinds of dependency problems without any benefit. http://lists.freebsd.org/pipermail/freebsd-ports/2011-January/065585.html Yuri pgpMukpVq0ajR.pgp Description: PGP signature
Re: archivers/unzip vs tar -xf
On Mon, Jun 06, 2011 at 11:50:36AM -0700, Xin LI wrote: On Mon, Jun 6, 2011 at 10:20 AM, Eir Nym eir...@gmail.com wrote: Since FreeBSD 5.4.0-Release libarchive(3) support to unpack zip files. Why not to use it to unpack distributives to workdir? I've downloaded all distros and tried to unpack them all with tar(1) with no problems. I use FreeBSD 9.0-CURRENT r222717. I think the reason is that at the beginning tar(1) does not work for a little fraction of distfiles, but yes, I think it's reasonable to take a step forward and add a new option to force using the info-zip implementation. We already have /usr/bin/unzip, built around the libarchive. You can check this (?, not sure) list archives as to why it isn't used by default. Yuri pgpbmmvzIeUSG.pgp Description: PGP signature
Re: Fping with WITH_IPV6=YES
On Fri, Jun 03, 2011 at 05:57:44PM +0400, Pavel Timofeev wrote: Is it normal that /usr/ports/net/fping doesn't work correct if you have WITH_IPV6=YES in /etc/make.conf? Port's Makefile has the following comment: # IPv6 not enabled by default, as it does not work together with IPv4. and looking at the code, it seems that fping can use either ipv6 or ipv4 but not both (see the #if(n)def's in the code). With -DWITH_IPV6: $ fping 192.168.1.4 192.168.1.252 fping: hostname nor servname provided, or not known $ fping 2001:470:28:4ba:89c1:323b:9032:86ea \ 2001:470:28:4ba:4a5b:39ff:feb3:b824 2001:470:28:4ba:89c1:323b:9032:86ea is alive 2001:470:28:4ba:4a5b:39ff:feb3:b824 is alive Without -DWITH_IPV6: $ fping 192.168.1.4 192.168.1.252 192.168.1.4 is alive 192.168.1.252 is alive $ fping 2001:470:28:4ba:89c1:323b:9032:86ea \ 2001:470:28:4ba:4a5b:39ff:feb3:b824 2001:470:28:4ba:89c1:323b:9032:86ea address not found 2001:470:28:4ba:4a5b:39ff:feb3:b824 address not found HTH, Yuri pgpH05Me42iRt.pgp Description: PGP signature
Re: 8-STABLE /usr/include/utmp.h and tmpx
On Fri, Jun 03, 2011 at 09:16:58PM -0400, Aryeh Friedman wrote: No will do even though I don't think I have a complete enough list of ports to make a proper report (if in fact it is a per port solution vs. fixing base) I don't see any related commits to 8-STABLE, which commit are you reffering to (and guessing on previous thread, such change isn't going to be MFCed)? Yuri pgpxIOfILnKWs.pgp Description: PGP signature
Re: FreeBSD Port: devel/apr1 (1.4.4 breaks apache22)
On Tue, May 17, 2011 at 02:17:29PM -0400, Mike Jakubik wrote: Hello, I just updated my apache22 from 2.2.17 to 2.2.18, this update also updated apr1 from 1.4.2 to 1.4.4. After restarting my web server i was not able to connect to anything. The logs did not show anything wrong but apache was taking up 100% cpu on all processes and the system load was steadily going up. Reverting back to apache 2.2.17 did not help, only solution was to downgrade apr1 to 1.4.2. I am running this on freebsd 7.4. Just for the record: apr-ipv6-devrandom-db48-1.4.4.1.3.11 apache-2.2.18 seem to work fine for me on 8.2-RELEASE/amd64 (Apache/2.2.18 (FreeBSD) mod_ssl/2.2.18 OpenSSL/0.9.8q DAV/2 SVN/1.6.16 Server). Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: Call for Testers: VirtualBox 4.0.6
I think 4.0.6 is quite stable now. Is it really less stable than the current version in ports 3.2.12? Why don't you at least put them into ports/emulators repository with -devel suffix like many other ports do? So that there is no need to go through the third party repository. Yuri ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org
Re: [CFT] A new mplayer and mencoder
On Fri, Mar 18, 2011 at 07:26:07PM +0100, Thomas Zander wrote: Dear all, I have prepared a recent snapshot for a an upcoming update for the mplayer and mencoder ports. You can find the tarball here: http://www.rrr.de/~riggs/mplayer/m20110318.tar.bz2 Please test the things that you usually do with it, I'd appreciate feedback! Thanks in advance, Riggs Hi Thomas, Everything seems to work fine, with few small issues (please see attached diff): - configure flags to enable/disable amr{n,w}b are spelled incorrectly: ./configure --help | grep -E 'amrnb|amrwb' --disable-libopencore_amrnb disable libopencore_amr narrowband --disable-libopencore_amrwb disable libopencore_amr wideband - the build is heavily (ab)using 'cc' despite CC is set to clang (even with --cc=${CC} passed to configure). There's another small fix in the diff to fix build failure where clang is actually used. HTH, Yuri diff -ruN mplayer.old/Makefile mplayer/Makefile --- mplayer.old/Makefile 2011-03-18 19:29:39.0 +0300 +++ mplayer/Makefile 2011-03-25 08:12:26.323326869 +0300 @@ -89,7 +89,9 @@ --disable-libdirac-lavc \ --disable-mencoder \ --disable-mpg123 \ - --disable-musepack + --disable-musepack \ + --cc=${CC} \ + --host-cc=${CC} .include ${.CURDIR}/Makefile.options diff -ruN mplayer.old/Makefile.options mplayer/Makefile.options --- mplayer.old/Makefile.options 2011-03-18 18:32:51.0 +0300 +++ mplayer/Makefile.options 2011-03-25 07:37:24.654326280 +0300 @@ -212,15 +212,15 @@ .endif .ifdef(WITH_AMR_NB) -CONFIGURE_ARGS+= --enable-libopencore-amrnb --enable-version3 +CONFIGURE_ARGS+= --enable-libopencore_amrnb --enable-version3 .else -CONFIGURE_ARGS+= --disable-libopencore-amrnb +CONFIGURE_ARGS+= --disable-libopencore_amrnb .endif .ifdef(WITH_AMR_WB) -CONFIGURE_ARGS+= --enable-libopencore-amrwb --enable-version3 +CONFIGURE_ARGS+= --enable-libopencore_amrwb --enable-version3 .else -CONFIGURE_ARGS+= --disable-libopencore-amrwb +CONFIGURE_ARGS+= --disable-libopencore_amrwb .endif .ifdef(WITH_GSM) diff -ruN mplayer.old/files/patch-ffmpeg__libavcodec__x86__snowdsp_mmx.c mplayer/files/patch-ffmpeg__libavcodec__x86__snowdsp_mmx.c --- mplayer.old/files/patch-ffmpeg__libavcodec__x86__snowdsp_mmx.c 1970-01-01 03:00:00.0 +0300 +++ mplayer/files/patch-ffmpeg__libavcodec__x86__snowdsp_mmx.c 2011-03-24 23:43:47.498333114 +0300 @@ -0,0 +1,19 @@ +--- ffmpeg/libavcodec/x86/snowdsp_mmx.c.orig 2011-03-24 23:27:25.485338703 +0300 ffmpeg/libavcodec/x86/snowdsp_mmx.c 2011-03-24 23:43:04.159326606 +0300 +@@ -675,14 +675,14 @@ + + #define snow_inner_add_yblock_sse2_end_8\ + sal $1, %%REG_c \n\t\ +- add $PTR_SIZE*2, %1 \n\t\ ++ addb $PTR_SIZE*2, %1 \n\t\ + snow_inner_add_yblock_sse2_end_common1\ + sar $1, %%REG_c \n\t\ + sub $2, %2 \n\t\ + snow_inner_add_yblock_sse2_end_common2 + + #define snow_inner_add_yblock_sse2_end_16\ +- add $PTR_SIZE*1, %1 \n\t\ ++ addw $PTR_SIZE*1, %1 \n\t\ + snow_inner_add_yblock_sse2_end_common1\ + dec %2 \n\t\ + snow_inner_add_yblock_sse2_end_common2 ___ freebsd-ports@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-ports To unsubscribe, send any mail to freebsd-ports-unsubscr...@freebsd.org