Created attachment 200244 [details] Update patch - Update to the latest version. Changelog: http://www.dhondt.de/new_calc.htm - Attempt to make the build reproducible concerning the date in the executable. - Drop (perhaps temporarily) the taucs dependency. It brings a secondary dependency on metis4 which we are not packaging. - Pet portlint.
Created attachment 200247 [details] Update calculix-ccx to version 2.15 Oops .. wrong patch.
Reporter is maintainer, is committer, assign accordingly, pending QA and ports approval.
(In reply to Kubilay Kocak from comment #2) I am afraid that assigning port PRs to me won't help get the port committed as it will a lot less visible to ports committers ... I now have a bunch of ports, that no ports committer is looking at because they are already assigned to a committer(?).
A note: without consulting me the dependency to arpack95 was removed and replaced with arpack-ng which is a parallel version of the same code. When I ported this years ago, there was an application note that the code depended on the serial version of arpack95, which also used an older version of lapack2. We did some adjustments to ensure the port worked correctly for CalculiX. I have no idea if CalculiX still works: I am consultimg upstream. Depending on what upstream says I may end up building with the internal arpack, which is a step backwards IMHO. PS. I shouldn't be assigned to this port, as I am not a ports committer so someone else has to approve it.
I got a word from the arpack people that the old arpack is unsafe to use, and arpack-ng should be strongly preferred. This is why I removed it. Yuri
It fails to fetch: => Attempting to fetch http://www.dhondt.de/ccx_2.15.src.tar.bz2 fetch: http://www.dhondt.de/ccx_2.15.src.tar.bz2: size mismatch: expected 1219901, actual 1434440
(In reply to Yuri Victorovich from comment #5) I have done some few tests now and the results appear to match what is expected. Still today the compilation instructions point to the original arpack website so if you had asked me I would have contacted the CalculiX developers first. Such changes are delicate and can break completely the known behaviour of such programs.
(In reply to Yuri Victorovich from comment #6) The authors are known to upload last minute fixes without changing the version :(. I will recalculate the checksums.
(In reply to Pedro F. Giffuni from comment #8) And the new version is now broken: ... gfortran8 -O -fpic -Wl,-rpath=/usr/local/lib/gcc8 -O3 -c keystart.f gfortran8 -O -fpic -Wl,-rpath=/usr/local/lib/gcc8 -O3 -c knotmpc.f gfortran8 -O -fpic -Wl,-rpath=/usr/local/lib/gcc8 -O3 -c labyrinth.f gfortran8 -O -fpic -Wl,-rpath=/usr/local/lib/gcc8 -O3 -c lab_straight_ppkrit.f gmake[2]: *** No rule to make target 'lbfgsb.o', needed by 'ccx_2.15.a'. Stop. gmake[2]: *** Waiting for unfinished jobs.... gmake[2]: Leaving directory '/usr/ports/cad/calculix-ccx/work/CalculiX/ccx_2.15/src' Lets wait a couple of days for upstream to fix it again!
Created attachment 200975 [details] Update calculix-ccx to version 2.15 Upsteam has rerolled the tarball, yet again, to fix the build issue.
Is the upstream issue now solved?
(In reply to kunda from comment #11) Yes. I am not a ports committer though so we have to wait.
(In reply to Pedro F. Giffuni from comment #12) Any sign of a ports committer ?
(In reply to kunda from comment #13) I did my part ... we need to bug the ports guys now.
Moin moin It unfortunately fails to build for me on current [...] gcc8 -O2 -pipe -fpic -fstack-protector -Wl,-rpath=/usr/local/lib/gcc8 -fno-strict-aliasing -O3 -I/usr/local/include -I/usr/local/include/spooles -DARCH="FreeBSD" -DSPOOLES -DARPACK -DMATRIXSTORAGE -DNETWORKOUT -c writeheading.c gmake[1]: *** No rule to make target '-lmetis', needed by 'ccx_2.15'. Stop. gmake[1]: *** Waiting for unfinished jobs.... In file included from nonlingeo.c:21: nonlingeo.c: In function 'nonlingeo': [...] mfg Tobias
(In reply to Tobias C. Berner from comment #15) Looks like you might be missing a dependency on math/metis.
(In reply to Tobias C. Berner from comment #16) Actually the issue is actually that I left -lmetis in the build Makefile. I decided to drop taucs (and metis) dependency until the deprecation of metis4 is resolved. I will update the patch soon. Thanks!
Created attachment 202059 [details] Updated diff (without metis issue) Sorry for that .. should be fixed (builds here).
*** Bug 236670 has been marked as a duplicate of this bug. ***
Thanks for effort.
Take now that I am ports committer.
A commit references this bug: Author: pfg Date: Thu Apr 18 15:33:30 UTC 2019 New revision: 499275 URL: https://svnweb.freebsd.org/changeset/ports/499275 Log: cad/calculix-ccx: Update to version 2.15. - Update to the latest version. Changelog: http://www.dhondt.de/new_calc.htm - Attempt to make the build reproducible concerning the date in the executable. - Drop temporarily) the taucs dependency. It brings a secondary dependency on metis4 which we are not packaging. We will bring taucs back when it depends on metis5 - Pet portlint. PR: 234146 Approved by: thierry (mentor) Changes: head/cad/calculix-ccx/Makefile head/cad/calculix-ccx/distinfo head/cad/calculix-ccx/files/patch-CalculiX.h head/cad/calculix-ccx/files/patch-Makefile head/cad/calculix-ccx/files/patch-date.pl head/cad/calculix-ccx/pkg-plist