Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 Conrad Meyer <konrad(a)tylerc.org&gt; changed: What |Removed |Added ---------------------------------------------------------------------------- Depends on| |476329 --- Comment #2 from Conrad Meyer <konrad(a)tylerc.org&gt; 2008-12-13 05:50:31 EDT --- Obviously (from scratch build) this depends on cudd. Adding blocker bug. -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 --- Comment #3 from Jason Tibbitts <tibbs(a)math.uh.edu&gt; 2009-03-09 00:54:07 EDT --- This failed to build for me; it looks like it's missing a build dependency on scons: + scons install PREFIX=/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.x86_64/usr EPREFIX=/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.x86_64/usr/bin INSTALLDIR=/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.x86_64/usr/share/python-polybori DOCDIR=/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.x86_64/usr/share/doc/python-polybori MANDIR=/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.x86_64/usr/share/man PYINSTALLPREFIX=/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.x86_64/usr/lib64/python2.6/site-packages DEVEL_PREFIX=/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.x86_64/usr 'CFLAGS=-O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m64 -mtune=generic -I/usr/include/cudd -I/usr/include/m4ri' 'CCFLAGS=-O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m64 -mtune=generic -I/usr/include/cudd -I/usr/include/m4ri' 'CXXFLAGS=-O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m64 -mtune=generic -I/usr/include/cudd -I/usr/include/m4ri' /var/tmp/rpm-tmp.TvllXs: line 30: scons: command not found I added scons as a build dep and the build fails later: In file included from /usr/include/boost/dynamic_bitset.hpp:16, from groebner/src/groebner_alg.h:16, from PyPolyBoRi/Poly_wrapper.cc:14: /usr/include/boost/dynamic_bitset/dynamic_bitset.hpp: In member function 'size_t boost::dynamic_bitset<Block, Allocator>::count() const': /usr/include/boost/dynamic_bitset/dynamic_bitset.hpp:1021: error: 'mode' cannot appear in a constant-expression /usr/include/boost/dynamic_bitset/dynamic_bitset.hpp:1021: error: template argument 1 is invalid /usr/include/boost/dynamic_bitset/dynamic_bitset.hpp:1021: error: expected '>' before '*' token /usr/include/boost/dynamic_bitset/dynamic_bitset.hpp:1021: error: expected '(' before '*' token /usr/include/boost/dynamic_bitset/dynamic_bitset.hpp:1021: error: expected primary-expression before '>' token No idea what's up there. -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 --- Comment #5 from Conrad Meyer <konrad(a)tylerc.org&gt; 2009-03-20 03:05:54 EDT --- So in mock on F-10 it errors because it can't link against cudd (probably because we haven't built cudd for F-10 (or at least haven't pushed the update)). In F-11 on koji it errors because ... boost is broken? -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 Mamoru Tasaka <mtasaka(a)ioa.s.u-tokyo.ac.jp&gt; changed: What |Removed |Added ---------------------------------------------------------------------------- Depends on| |491537 -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 Mamoru Tasaka <mtasaka(a)ioa.s.u-tokyo.ac.jp&gt; changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |mtasaka(a)ioa.s.u-tokyo.ac.jp Status Whiteboard|NotReady | --- Comment #7 from Mamoru Tasaka <mtasaka(a)ioa.s.u-tokyo.ac.jp&gt; 2009-03-23 16:13:32 EDT --- Comments just for build issue: ( I have only checked if this package builds or not on dist-f11 and have not tried to review other things on this package) ! boost issue is now fixed (workaround applied) (bug 491537) For this package: - BR: scons is needed - Still does not build because this package does not pass check-buildroot: ----------------------------------------------------------------- + /usr/lib/rpm/check-buildroot /builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.1.x86_64/usr/share/python-polybori/ipbori/ipythonrc-polybori:execute pyroot = '/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.1.x86_64/usr/lib64/python2.6/site-packages' Found '/builddir/build/BUILDROOT/python-polybori-0.5-1.fc11.1.x86_64' in installed files; aborting error: Bad exit status from /var/tmp/rpm-tmp.yxVT94 (%install) ------------------------------------------------------------------ At least ------------------------------------------------------------------ sed -i -e 's|%{buildroot}||' %{buildroot}%{_datadir}/%{name}/ipbori/ipythonrc-polybori ------------------------------------------------------------------ is needed to fix this issue - After fixing these two, this package builds on dist-f11: http://koji.fedoraproject.org/koji/taskinfo?taskID=1255206 Again I just checked if this package builds or not. -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 --- Comment #9 from Conrad Meyer <konrad(a)tylerc.org&gt; 2009-03-23 16:28:27 EDT --- Except, I'm not sure how to do that -- can you make a patch? -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 --- Comment #11 from Conrad Meyer <konrad(a)tylerc.org&gt; 2009-03-23 23:16:24 EDT --- Thanks. http://konradm.fedorapeople.org/fedora/SPECS/python-polybori.spec http://konradm.fedorapeople.org/fedora/SRPMS/python-polybori-0.5-3.fc10.s... -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 --- Comment #13 from Mamoru Tasaka <mtasaka(a)ioa.s.u-tokyo.ac.jp&gt; 2009-03-24 14:37:22 EDT --- Some notes: * %define -> %global - Current packaging guidelines suggest to use %global instead of %define, especially when defining nested macros, see: https://fedoraproject.org/wiki/Packaging/Python#System_Architecture https://fedoraproject.org/wiki/PackagingDrafts/global_preferred_over_define (The latter one is still under "Drafts" category, but this is already accepted by FESCo) * %debug_package ----------------------------------------------------------------- # and apparently # also stops checking other libs once it hits the first static lib ----------------------------------------------------------------- - The problem is not there. The real issue is that the installed shared libraries are already stripped: ----------------------------------------------------------------- Symlinking from libgroebner-0.5.0.so to /builddir/build/BUILDROOT/python-polybori-0.5-3.fc11.1.x86_64/usr/lib64/libgroebner.so g++ -o polybori/libpolybori-0.5.0.so.0.0.0 -s -shared -Wl,-soname,libpolybori-0.5.0.so.0 .... ----------------------------------------------------------------- note that "-s" option is passed to g++. Perhaps modifying "LDFLAGS_LINUX" in polybori/Makefile.in will solve this issue. * License - Seems GPLv2+. Would you check this? * Requires - Would you check the dependency "i.e. Requires" for -devel subpackage? ! For example, polybori/groebner/groebner_alg.h contains: ----------------------------------------------------------------- 12 #include <polybori.h> 13 #include "groebner_defs.h" 14 #include "pairs.h" 15 #include <boost/dynamic_bitset.hpp> ----------------------------------------------------------------- This means -devel subpackage should have "Requires: boost-devel" (here I am not saying about BuildRequires). ----------------------------------------------------------------- # rpm -ql python-polybori-devel | xargs grep -h 'include ' | sort | uniq ----------------------------------------------------------------- will show some useful information. ! And this output shows that polybori/CDDManager.h contains: ----------------------------------------------------------------- 102 #define CDDManager_h_ 103 #include "cacheopts.h" 104 // load basic definitions 105 #include "pbori_defs.h" ----------------------------------------------------------------- Would you check if cacheopts.h can be really removed? - Also please check the dependency for main package. ! For example, polybori/nf.py contains: ----------------------------------------------------------------- 81 global mat_counter 82 mat_counter=mat_counter+1 83 import Image, ImageDraw ----------------------------------------------------------------- This may mean that python-polybori should have "Requires: python-imaging". ----------------------------------------------------------------- # rpm -ql python-polybori | grep -v /usr/share/doc | LANG=C xargs grep -h 'import ' | grep -v Binary | sort | uniq ----------------------------------------------------------------- will show some useful information. ! Also ----------------------------------------------------------------- $ LANG=C ipbori /usr/bin/ipbori: line 66: ipython: command not found ----------------------------------------------------------------- Perhaps "Requires: ipython" is needed. * SourceURL - Please follow: https://fedoraproject.org/wiki/Packaging/SourceURL#Sourceforge.net * Linkage on installed system libraries ------------------------------------------------------------------ # rpmlint python-polybori python-polybori.i586: W: undefined-non-weak-symbol /usr/lib/libpolybori-0.5.0.so.0.0.0 Cudd_zddUnion python-polybori.i586: W: undefined-non-weak-symbol /usr/lib/libpolybori-0.5.0.so.0.0.0 _Z12defaultErrorSs python-polybori.i586: W: undefined-non-weak-symbol /usr/lib/libpolybori-0.5.0.so.0.0.0 Cudd_ReadZddSize python-polybori.i586: W: undefined-non-weak-symbol /usr/lib/libpolybori-0.5.0.so.0.0.0 cuddCacheInsert1 ..... ..... python-polybori.i586: W: undefined-non-weak-symbol /usr/lib/libgroebner-0.5.0.so.0.0.0 Cudd_zddUnion python-polybori.i586: W: undefined-non-weak-symbol /usr/lib/libgroebner-0.5.0.so.0.0.0 _Z12defaultErrorSs python-polybori.i586: W: undefined-non-weak-symbol /usr/lib/libgroebner-0.5.0.so.0.0.0 Cudd_ReadZddSize ..... ..... ------------------------------------------------------------------ - These two libraries undefined non-weak symbols. This cannot be allowed when these libraries also have in -devel subpackage named "libfoo.so" used for linkage against these libraries, because these symbols will cause linkage error. * Documents - I prefer to include README as %doc of main package for this case. - ChangeLog seems useful for -devel subpackage. -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 --- Comment #15 from Mamoru Tasaka <mtasaka(a)ioa.s.u-tokyo.ac.jp&gt; 2009-03-25 18:20:27 EDT --- Well, today I had no time to review your newest srpm, however: (In reply to comment #14) - What do you mean by "This doesn't fix it"? Your latest srpm now correctly generates debuginfo rpm (note that of course static archives are ignored) - Please check if only "boost-devel" is needed as Requires. For example it seems that "Requires: cudd-devel" is also needed (as polybori/pbori_algo.h contains ------------------------------------------------------- 162 // temporarily 163 #include "cudd.h" 164 #include "cuddInt.h" 165 #include "CCuddInterface.h" ------------------------------------------------------- ) (However m4ri-devel doesn't seem to be needed for Requires) - Please remove these undefined non-weak symbols by making those two libraies linked against proper libraries. Most of these undefined symbols on these 2 libraies are actually in libcudd.so.2, for making libpolybori.so and libgroebner.so linked against libcudd.so.2 (by adding "-lcudd" to LDFLAGS or so, for example) will solve _most_ of the problems, but as far as I checked not all. The trouble is that the symbol "_Z12defaultErrorSs" (after demangled it is "defaultError(std::string)") is not defined anywhere. Actually obj/cuddObj.cc in cudd-2.4.1 seems to define this funtion, however build.log for cudd shows that this source is not compiled and so is not included in the libraries in cudd binary rpms. So currently no libraries contain "Z12defaultErrorSs" symbol and this will cause linkage error I guess the code (or Makefiles) in cudd must need fixing so that defaultError(std::string) symbols is actually included in the libraries in cudd. -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 --- Comment #17 from Conrad Meyer <konrad(a)tylerc.org&gt; 2009-03-28 22:14:39 EDT --- I don't get warnings from rpmlint about undefined non-weak symbols other than _Z12defaultErrorSs so I think I did it right: http://konradm.fedorapeople.org/fedora/SPECS/python-polybori.spec http://konradm.fedorapeople.org/fedora/SRPMS/python-polybori-0.5-5.fc10.s... Now to fix cudd, I guess. -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 --- Comment #19 from Mamoru Tasaka <mtasaka(a)ioa.s.u-tokyo.ac.jp&gt; 2009-03-29 13:06:39 EDT --- Created an attachment (id=337176) --> (https://bugzilla.redhat.com/attachment.cgi?id=337176) Patch to build libcuddobj.so Well, - python-polybori side now seems good (I have not retried full review yet) - however cudd side is not yet good, as libpolybori.so cannot resolve defaultError(std::string) symbol: ----------------------------------------------------- # ldd -r /usr/lib/libpolybori-0.5.0.so undefined symbol: _Z12defaultErrorSs (/usr/lib/libpolybori-0.5.0.so) linux-gate.so.1 => (0x00e7d000) libcudd.so.2 => /usr/lib/libcudd.so.2 (0x00f24000) libm4ri-0.0.20081029.so => /usr/lib/libm4ri-0.0.20081029.so (0x0050c000) libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x006b8000) libm.so.6 => /lib/libm.so.6 (0x0057e000) libgcc_s.so.1 => /lib/libgcc_s.so.1 (0x005f0000) libc.so.6 => /lib/libc.so.6 (0x00110000) libmtr.so.2 => /usr/lib/libmtr.so.2 (0x00281000) libcuddst.so.2 => /usr/lib/libcuddst.so.2 (0x00abf000) libcuddutil.so.2 => /usr/lib/libcuddutil.so.2 (0x00b9a000) libepd.so.2 => /usr/lib/libepd.so.2 (0x00c75000) /lib/ld-linux.so.2 (0x00b76000) ------------------------------------------------------- As I said in the previous comment defaultError(std::string) is defined in obj/cuddObj.cc - A extra library under obj needs building. This library is not built by default so a patch is needed. Your patch builds a static archive under obj/ but does not seem to build shared library under obj/ . ! By the way the library under obj/ directory is named as libobj.so, however I think this naming is very bad and this library should be renamed as libcuddobj.so . ! Also without a path the rebuilt libcuddobj.so (after renamed) library is broken as this contains undefined non-weak symbols :(, because - although this is C++ library, "cc" is used for linkage. Note that libcudd.so is C library - this library is not properly linkaged against libcudd.so For me libcuddobj.so can be properly built by the attached patch. - After this, libpolybori.so and libgroebner.so should be linkaged against both libcudd.so and libcuddobj.so. ! Note First I tried to merge libcuddobj.so into libcudd.so, however as said before libcudd.so is C library but libcuddobj.so contains C++ function, so this should not be done. -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 Mamoru Tasaka <mtasaka(a)ioa.s.u-tokyo.ac.jp&gt; changed: What |Removed |Added ---------------------------------------------------------------------------- Flag|fedora-review? |fedora-review+ --- Comment #21 from Mamoru Tasaka <mtasaka(a)ioa.s.u-tokyo.ac.jp&gt; 2009-03-30 12:43:22 EDT --- Okay, now it seems okay with replacing like below: ----------------------------------------------------------- sed -i -e 's|^LDFLAGS_LINUX.*-s|LDFLAGS_LINUX = -lcudd -lcuddobj|' \ polybori/Makefile.in sed -i -e "s|\['-s'\]|['-lcudd -lcuddobj']|" SConstruct ----------------------------------------------------------- http://koji.fedoraproject.org/koji/taskinfo?taskID=1265939 ----------------------------------------------------------- This package (python-polybori) is APPROVED by mtasaka ----------------------------------------------------------- -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.

Please do not reply directly to this email. All additional comments should be made in the comments box of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=476346 Dennis Gilmore <dennis(a)ausil.us&gt; changed: What |Removed |Added ---------------------------------------------------------------------------- Flag|fedora-cvs? |fedora-cvs+ --- Comment #23 from Dennis Gilmore <dennis(a)ausil.us&gt; 2009-04-01 12:38:43 EDT --- CVS Done -- Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.