Please do not reply directly to this email. All additional comments should be made in the comments box of this bug.
Summary: Review Request: python-polybori - Framework for Boolean Rings
https://bugzilla.redhat.com/show_bug.cgi?id=476346
Summary: Review Request: python-polybori - Framework for Boolean Rings Product: Fedora Version: rawhide Platform: All OS/Version: Linux Status: NEW Severity: medium Priority: medium Component: Package Review AssignedTo: nobody@fedoraproject.org ReportedBy: konrad@tylerc.org QAContact: extras-qa@fedoraproject.org CC: notting@redhat.com, fedora-package-review@redhat.com Estimated Hours: 0.0 Classification: Fedora
Spec URL: http://konradm.fedorapeople.org/fedora/SPECS/python-polybori.spec SRPM URL: http://konradm.fedorapeople.org/fedora/SRPMS/python-polybori-0.5-1.fc9.src.r... Description: PolyBoRi is a special purpose computer algebra system for computations in Boolean Rings. The core is a C++ library, which provides high-level data types for Boolean polynomials and related structures. As a unique approach, binary decision diagrams are used as internal storage type for polynomial structures. On top of this, we provide a Python interface for parsing of complex polynomial systems, as well as for sophisticated and extendable strategies for Groebner base computation.
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 #1 from Conrad Meyer konrad@tylerc.org 2008-12-13 05:48:57 EDT --- Koji scratch build:
http://koji.fedoraproject.org/koji/taskinfo?taskID=996510
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@tylerc.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Depends on| |476329
--- Comment #2 from Conrad Meyer konrad@tylerc.org 2008-12-13 05:50:31 EDT --- Obviously (from scratch build) this depends on cudd. Adding blocker 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
Bug 476346 depends on bug 476329, which changed state.
Bug 476329 Summary: Review Request: cudd - CU Decision Diagram Package https://bugzilla.redhat.com/show_bug.cgi?id=476329
What |Old Value |New Value ---------------------------------------------------------------------------- Status|NEW |ASSIGNED Resolution| |RAWHIDE Status|ASSIGNED |CLOSED
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@math.uh.edu 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.
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 #4 from Conrad Meyer konrad@tylerc.org 2009-03-19 00:07:06 EDT --- I'm having trouble getting PolyBoRi to link against cudd for whatever reason. This will take some work.
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@tylerc.org 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?
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
Jason Tibbitts tibbs@math.uh.edu changed:
What |Removed |Added ---------------------------------------------------------------------------- Status Whiteboard| |NotReady
--- Comment #6 from Jason Tibbitts tibbs@math.uh.edu 2009-03-20 14:15:33 EDT --- Please clear the whiteboard when someone gets this building.
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@ioa.s.u-tokyo.ac.jp changed:
What |Removed |Added ---------------------------------------------------------------------------- Depends on| |491537
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
Bug 476346 depends on bug 491537, which changed state.
Bug 491537 Summary: dynamic_bitset/dynamic_bitset.hpp won't compile https://bugzilla.redhat.com/show_bug.cgi?id=491537
What |Old Value |New Value ---------------------------------------------------------------------------- Status|NEW |MODIFIED Resolution| |RAWHIDE Status|MODIFIED |CLOSED
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@ioa.s.u-tokyo.ac.jp changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |mtasaka@ioa.s.u-tokyo.ac.jp Status Whiteboard|NotReady |
--- Comment #7 from Mamoru Tasaka mtasaka@ioa.s.u-tokyo.ac.jp 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.
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 #8 from Conrad Meyer konrad@tylerc.org 2009-03-23 16:27:06 EDT --- Thanks! I will grab the .spec from the Koji task.
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@tylerc.org 2009-03-23 16:28:27 EDT --- Except, I'm not sure how to do that -- can you make a patch?
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 #10 from Mamoru Tasaka mtasaka@ioa.s.u-tokyo.ac.jp 2009-03-23 23:06:01 EDT --- (In reply to comment #9)
Except, I'm not sure how to do that -- can you make a patch?
My srpm can be downloaded from: http://koji.fedoraproject.org/scratch/mtasaka/task_1255206/
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@tylerc.org 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.src....
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@ioa.s.u-tokyo.ac.jp changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |ASSIGNED AssignedTo|nobody@fedoraproject.org |mtasaka@ioa.s.u-tokyo.ac.jp Flag| |fedora-review?
--- Comment #12 from Mamoru Tasaka mtasaka@ioa.s.u-tokyo.ac.jp 2009-03-24 14:04:22 EDT --- Assigning.
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@ioa.s.u-tokyo.ac.jp 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.
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 #14 from Conrad Meyer konrad@tylerc.org 2009-03-24 16:07:29 EDT --- (In reply to comment #13)
- %define -> %global
Done.
- %debug_package Perhaps modifying "LDFLAGS_LINUX" in polybori/Makefile.in will solve this issue.
This doesn't fix it --- I also removed -s as a link flag from SConstruct, and now it builds without stripping at link time.
- License
- Seems GPLv2+. Would you check this?
Ah. The only source I could find (sourceforge.net) just said GPL, so I assumed GPLv3. I guess LICENSE says GPLv2+, good (and at least one source file says the license information is in LICENSE).
- Requires This means -devel subpackage should have "Requires: boost-devel" (here I am not saying about BuildRequires).
Good catch; fixed.
# 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?
Well, it's an empty file and rpm complains about that. I will not remove it, for now.
- Also please check the dependency for main package. This may mean that python-polybori should have "Requires:
python-imaging".
Done, thanks for spotting that.
$ LANG=C ipbori /usr/bin/ipbori: line 66: ipython: command not found
Perhaps "Requires: ipython" is needed.
Yep.
- SourceURL
Thanks, I couldn't find that when I was originally packaging this. Fixed.
- Linkage on installed system libraries
- 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.
What do I need to fix for this?
- Documents
- I prefer to include README as %doc of main package for this case.
- ChangeLog seems useful for -devel subpackage.
Ok, added.
Update URLs: http://konradm.fedorapeople.org/fedora/SPECS/python-polybori.spec http://konradm.fedorapeople.org/fedora/SRPMS/python-polybori-0.5-4.fc10.src....
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@ioa.s.u-tokyo.ac.jp 2009-03-25 18:20:27 EDT --- Well, today I had no time to review your newest srpm, however:
(In reply to comment #14)
- %debug_package Perhaps modifying "LDFLAGS_LINUX" in polybori/Makefile.in will solve this issue.
This doesn't fix it --- I also removed -s as a link flag from SConstruct, and now it builds without stripping at link time.
- 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)
- Requires This means -devel subpackage should have "Requires: boost-devel" (here I am not saying about BuildRequires).
Good catch; fixed.
- 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)
- Linkage on installed system libraries
- 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.
What do I need to fix for this?
- 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.
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 #16 from Conrad Meyer konrad@tylerc.org 2009-03-25 19:16:22 EDT --- (In reply to comment #15)
- 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)
Sorry, that alone didn't fix it. I also had to make changes via sed to the SConstruct file.
- 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 (However m4ri-devel doesn't seem to be needed for Requires)
Ah, yes, this seems correct.
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.
Ok, I will fix this to link against cudd at build time, and fix cudd to include that symbol. I wasn't sure if libraries could link to other libraries (I am fairly new to doing anything beyond basic linking).
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@tylerc.org 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.src....
Now to fix cudd, I guess.
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 #18 from Conrad Meyer konrad@tylerc.org 2009-03-28 22:57:13 EDT --- I think I fixed it with the new patch here: http://koji.fedoraproject.org/koji/taskinfo?taskID=1263843
Though I did not test it yet.
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@ioa.s.u-tokyo.ac.jp 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.
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 #20 from Conrad Meyer konrad@tylerc.org 2009-03-29 17:41:43 EDT --- Ok, built:
http://koji.fedoraproject.org/koji/taskinfo?taskID=1264624
Time to fix polybori to link against libcuddobj as well :).
Thank you very much for the patch.
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@ioa.s.u-tokyo.ac.jp changed:
What |Removed |Added ---------------------------------------------------------------------------- Flag|fedora-review? |fedora-review+
--- Comment #21 from Mamoru Tasaka mtasaka@ioa.s.u-tokyo.ac.jp 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 -----------------------------------------------------------
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@tylerc.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Flag| |fedora-cvs?
--- Comment #22 from Conrad Meyer konrad@tylerc.org 2009-03-31 17:54:01 EDT --- New Package CVS Request ======================= Package Name: python-polybori Short Description: Framework for Boolean Rings Owners: konradm Branches: F-10 InitialCC:
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@ausil.us changed:
What |Removed |Added ---------------------------------------------------------------------------- Flag|fedora-cvs? |fedora-cvs+
--- Comment #23 from Dennis Gilmore dennis@ausil.us 2009-04-01 12:38:43 EDT --- CVS Done
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@tylerc.org changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |CLOSED Resolution| |RAWHIDE
--- Comment #24 from Conrad Meyer konrad@tylerc.org 2009-04-03 21:04:25 EDT --- Imported and built in rawhide; closing.
http://koji.fedoraproject.org/koji/taskinfo?taskID=1276530
package-review@lists.fedoraproject.org
-
Red Hat Bugzilla