A few (mostly Fortran-based) ports need to be patched because GCC 10
is stricter. A handful of ports need to be restricted to GCC 9 because
they fail with 10.
5 ports still fail with GCC 10. However, seeing as this work has been
ongoing since 2020-05-24, it is simply time to make the commit and
notify the affected maintainers.
While here, pet portlint (Makevar order).
PR: 246700
Submitted by: gerald
Approved by: antoine (after many, many, -exp runs)
- Disable Perl bindings for now (there are issues)
- Add missing bits for the proper Python support
- Install examples (and put them where they belong)
The graphics/py-pillow6 port was removed in r559976 as it was for
Python 2.7 only and no longer referenced in the default cases.
However it was still defined in Mk/Uses/python.mk in PY_PILLOW,
leaving a dangling reference.
Remove it from Mk/Uses/python.mk, leaving only the Python 3
version of that port, graphics/py-pillow, defined and bump the
minimal verion of Python to 3.6 for affected ports.
"Patch release with various bug/sanitizer/security fixes, primarily related to
reading corrupted input files."
<https://github.com/AcademySoftwareFoundation/openexr/releases/tag/v2.5.4>
Cherry-pick upstream commit for GCC 11 compatibility:
<a40a6151c4>
Since this changes the patchlevel of the solibs, bump PORTREVISION of dependent
ports just to be sure.
(graphics/py-openshadinglanguage is unaltered and .includes the revision bump
from .../openshadinglanguage.)
MFH: 2021Q1 (this requires discussion by ports-secteam@!)
the package is still 4.7. as a proof, SHA256 in distinfo is identical.
confusingly, the version in "About" says "master", not 4.8 nor 4.7.
this PR should be reopened. (Tomoyuki Sakura)
Apparently I forgot to remove a line causing the wrong cura version
to be claimed as an update to 4.8
I think I have it right this time. Apologies I should have noticed.
I'll take the pointyhat and go sit in the corner.
PR: ports/251236
Submitted by: sko@rostwald.de
Reported by: trombik1973@gmail.com
Since r558913 Python 3.8 incorporates BPO-42604 [1] which changed the
shared libs naming scheme. This means "EXT_SUFFIX" is now derived from
SOABI and yields with Python 3.8 to ".cpython-38.so" instead of ".so".
The affected ports strip the libaries in the "post-install" target via
hardcoded path(s) and the build fails at the end because the new extension
is not expected at this place.
Remedy the issue by adding wildcards to these paths. This should also
prepare the ports for future Python releases, which will use the new shared
libs naming scheme.
[1] https://bugs.python.org/issue42604
PR: 252057
Reported by: John Kennedy
Reviewed by: fluffy, koobs
Approved by: koobs (python)
- Add some dependencies which might be detected by Find_Package but not registered;
- Bump PORTREVISION of consumers.
PR: 250833
Submitted by: /me
Approved by: tcberner@ (KDE / maintainer) + exp-run
This commit combines several updates.
- Update Qt to 5.12.2
- Update PyQt to 5.15.1
- Since PySide 5.15.1 is broken with Qt 5.15.2, so PySide and Shiboken are also updated to 5.15.2.
- Update sip to 5.4.0
SIP is a collection of tools to create Python bindings for C and C++ libraries and used by PyQt and wxPython. There are some changes with sip5 [1]:
- python 3.5+ is required
- sip drops support of old deprecated methods as sipdistutils & cie.
- this version breaks also PyQt5 extension ABI. SIP files will be installed in ${PYTHON_SITELIBDIR}/PyQt${_PYQT_VERSION}/bindings
- some ports will use the new pyqtbuilder package with the pyproject.toml setup. Instead if the project have a setup.py, you should use sip-build to build it.
- if a port needs sip, it should also needs pysip. BTW, py-qt5-core requires pysip, so that should be enough for PyQt packages.
List of ports impacted by this change. Most of the patches have been integrated or are in the process of being integrated upstream:
- devel/libsavitar
- graphics/py-python-poppler-qt5
- net-im/scudcloud
- net/libarcus
- print/py-frescobaldi
- science/py-veusz
- graphics/qgis and graphics/qgis-ltr
- deskutils/calibre
A special note regarding calibre. New versions require sip>=5, we update it to the latest version (thanks to madpilot@)
science/scidavis will remove the PyQt binding soon and there is no patch planned, so we remove the python option (ok makc@)
While here, convert some ports to USE_PYQT (cad/cura, cad/uranium, devel/eric6, ...)
Thanks to tcberner and adridg!
[1] https://www.riverbankcomputing.com/static/Docs/sip/
PR: 250853
Exp-run by: antoine
0.11 failed to produce dxf files which could be imported into AutoCAD.
This bugfix release improves DXF importing from 10% to about 90%.
Found out via: repology.org
This implies:
- french /aster builds fine with clang, but it seg-faults: switched to Gcc
- ASTK is upgraded to 2019.0-1
- math/mumps is upgraded to 5.1.2-aster7
- TFEL/MFront is now required, but when the compilers of tfel and aster are
different, it cannot be linked
. recopopy science/tfel to science/tfel-edf and depends on science/tfel-edf
. science/tfel-edf is built by Gcc, from the sources included in Aster's
tarball
. register the conflicts
Sync up with cura and cura-engine
fixes to Uranium
Tooltip areas should not steal mouse events
This prevents us from using the tooltip area in a pop-up where the Z
coordinates are not a good solution.
Round coordinates when sampling pixel for selection ray
Pixel coordinates must be integer for the Python bindings of Qt.
In Cura version 4.7 a bug regarding slowdowns in circular perimeters was
introduced (or at list gets triggered much more easily than in earlier
versions):
https://github.com/Ultimaker/Cura/issues/8321
This issue makes v4.7.x hardly usable.
Sticking to v4.5 has become quite "tricky" to say the least, due to
the massive python-induced dependency hell and lots of dependencies
being too new and/or incompatible for the 4.5 version.
Version 4.8 has been released on November 10 that fixes this glitch in
the simplification algorithm. The issue is still open, but is considered
to be solved with 4.8 and changed default settings for Maximum resolution
and Maximum Deviation.
From the Changelog of 4.8:
Fixed under-simplification (blobs, zits) on some printer models. An oversight
in 4.6.x resulted in an oversimplification (smoothing) of models.
The attempted fix in 4.7.x overcompensated, which gave difficulty
(zits, blobs) for some printer models when the resulting gcode became
too intricate. This is now fixed, though some profiles might need
to be updated, since they where made against 4.6.x, and therefore may
rely on the over-simplification.
PR: ports/251236
Submitted by: Sebastian <sko@rostwald.de>
- Add a pkg-message explaining how to access or sync datum grids from the CDN
(disabled by default)
- Add curl and tiff dependencies since proj can read datum grids from remote
COG (Cloud Optimized Geotiff)
- Take maintainer'ship (ok wen@)
PR: 245611
Submitted by: wen (first version)
Reviewed by: wen, rhurlin
Approved by: wen
Exp-run by: antoine
OpenCascade APIs changed significantly in some areas - FreeCAD
is affected on progress bars and some message output. While
messages are easily fixed, the affected progress bars have to
be enabled for now.
Patches adapted (some manual fuzzing required) from upstream
2cb9b147f1fd9cdb9de9
- switch to the more modern version of librsvg2 on architectures
supporting rust
- this will fix some graphical issues on these architectures
PR: 250276
Exp-run by: antoine
Submitted by: tobik
Differential Revision: https://reviews.freebsd.org/D18878
crashes when several JDK versions are installed and javawm(1) picks
higher version (happens when /usr/ports/Mk/bsd.java.mk is absent,
contrary to documented selection process which is designed to behave
almost identically WRT presence/absence of /usr/ports/Mk/bsd.java.mk
- While here, simplify the script a bit and remove keywords and their
expansion, they are useless for wrapper scripts
PR: 250268
Submitted by: rodrigo
- Update PyQt to 5.15.0, py-sip to 4.19.24 and qscintilla2 to 2.11.5
- Use pypi for PYQT [1]
- Add devel/py-qt5-sip (Don't use it for now. It will be used in future updates of PyQt with py-sip >= 5)
- Add missing py-qt5 ports: comms/py-qt5-sensors, devel/py-qt5-location, devel/py-qt5-remoteobjects, net/py-qt5-networkauth
- Strip libs
- Refactor *_PATH and *_PORT parts
[1] RiverBankComputing doesn't offer links to download latest version of some ports
PR: 247369
Reviewed by: tcberner
Differential Revision: https://reviews.freebsd.org/D25749
Exp-run by: antoine
Also, it adds math/py-numpy-stl - an optional dependency that makes
loading STL files much faster. Without it, Cura was complaining:
2020-09-20 02:19:31,877 - WARNING - [MainThread] STLReader.STLReader.<module> [32]: Could not find numpy-stl, falling back to slower code.
I'm not sure where the 6.x version in cad/fdm_materials came from, seems
like now that project has its version in sync with the Cura version,
so I chose to increment PORTEPOCH for that. [Greg]
- fdm_materials adds some new materials usable with Cura
- I had to bump PORTEPOCH on a couple of ports
- py-numpy-stl was added on a separate commit
- Bump version checking in Cura on some dependancies
[db]
PR: ports/249468
Submitted by: greg@unrelenting.technology
Extend the expiration of unmaintained leaf ports that are broken with
-fno-common and previously had a short expiration date. The new date is set
to 2021-04-01, meaning the ports will be in the 2021Q2 branch, giving people
until 2021-06-30 (10 months from now) to fix issues.
Improve the wording of the deprecation message, mentioning both llvm 11 and
-fno-common.
Bump portrevision to force reinstall. This way the warning message about
deprecation will be displayed to users installing or reinstalling the
software.
MFH: 2020Q3 (maybe)
Add -fcommon to cad/gmsh to fix the build on current with llvm 11.
This is a stop-gap measure since an update which fixes the -fno-common issue
is waiting for approval.
PR: 248996
CQ-editor is a CadQuery GUI editor based on PyQT.
Graphical debugger for CadQuery scripts
- Step through script and watch how your model changes
CadQuery object stack inspector
- Visual inspection of current workplane and selected items
- Insight into evolution of the model
Export to various formats
- STL
- STEP
WWW: https://github.com/CadQuery/CQ-editor/
PR: 248212
Submitted by: Neal Nelson <ports@nicandneal.net>
CadQuery is an intuitive, easy-to-use Python module for building parametric 3D
CAD models. Using CadQuery, you can write short, simple scripts that produce
high quality CAD models. It is easy to make many different objects using a
single script that can be customized.
WWW: https://github.com/CadQuery/cadquery
PR: 248210
Submitted by: Neal Nelson <ports@nicandneal.net>
OCP is a Python wrapper for OCCT (OpenCascade) generated using pywrap
primarily used by CadQuery.
WWW: https://github.com/CadQuery/OCP
PR: 248202
Submitted by: Neal Nelson <ports@nicandneal.net>
ilmbase and openexr were updated to v2.5.3 and their shlibs changed
from *.25.0.1 to *.25.0.2.
This is a PORTREVISION bump to ports using ilmbase or openexr
to chase that change and trigger rebuild.
Please refer to the 20200811 UPDATING entry when upgrading
dependent (*-emacs26-*) ports.
Port changes:
- depend on math/gmp
- match upstream by turning CAIRO, HARFBUZZ, and JSON options and on and
turning MAGICK off by default
- remove OPENMP check for graphics/ImageMagick as the openmp is now
included in base
- update EMACS_VER in Mk/Uses/emacs.mk
- bump USES=emacs ports or remove BROKEN for net-im/jabber.el and
deskutils/howm, which now build
Submitted by: HIROSE Yuuji <yuuji@gentei.org> (canna patch)
Reviewed by: ashish
Differential Revision: https://reviews.freebsd.org/D23966
A Python package to create and modify DXF drawings, independent from the DXF
version. You can open/save every DXF file without losing any content (except
comments), Unknown tags in the DXF file will be ignored but preserved for
saving. With this behavior it is possible to open also DXF drawings that
contains data from 3rd party applications.
PR: 248203
Submitted by: Neal Nelson
Approved by: tcberner (mentor)
Differential Revision: https://reviews.freebsd.org/D25890
Recent changes in gfortran seem to have broken this port so that it doesn't
produce valid results anymore. See [1] for more details.
This port would likely be a good candidate for LLVM Fortran but I currently
don't have time for to try (patches welcome).
Use dos2unix, as it seems necessary but not sufficient for a future fix.
Pet portlint while here.
Approved by: therry (mentor)
[1] https://github.com/nasa/NASTRAN-95/issues/15
/usr/local/bin/ld: /wrkdirs/usr/ports/cad/alliance/work/alliance/src/pat/src/.libs/libPat.a(pat_desc_y.o):(.bss+0x24): multiple definition of `pat_decl_y_errflag'; /wrkdirs/usr/ports/cad/alliance/work/alliance/src/pat/src/.libs/libPat.a(pat_decl_y.o):(.bss+0x20): first defined here
/usr/local/bin/ld: /wrkdirs/usr/ports/cad/alliance/work/alliance/src/pat/src/.libs/libPat.a(pat_desc_y.o):(.bss+0x10): multiple definition of `pat_decl_y_val'; /wrkdirs/usr/ports/cad/alliance/work/alliance/src/pat/src/.libs/libPat.a(pat_decl_y.o):(.bss+0x10): first defined here
/usr/local/bin/ld: /wrkdirs/usr/ports/cad/alliance/work/alliance/src/pat/src/.libs/libPat.a(pat_desc_y.o):(.bss+0x0): multiple definition of `pat_decl_y_lval'; /wrkdirs/usr/ports/cad/alliance/work/alliance/src/pat/src/.libs/libPat.a(pat_decl_y.o):(.bss+0x0): first defined here
/usr/local/bin/ld: /wrkdirs/usr/ports/cad/alliance/work/alliance/src/pat/src/.libs/libPat.a(pat_desc_y.o):(.bss+0x2c): multiple definition of `pat_decl_y_debug'; /wrkdirs/usr/ports/cad/alliance/work/alliance/src/pat/src/.libs/libPat.a(pat_decl_y.o):(.bss+0x24): first defined here
http://package18.nyi.freebsd.org/data/121amd64-default-PR244494/2020-06-10_18h04m44s/logs/errors/alliance-5.1.1_1.log
Also USES=compiler:c++0x & USE_GCC=any can be reduced to USE_GCC=yes
PR: 246700
for starting with 'USB' (which would actually be 'U' for us). I guess
somewhere it is only splitting on 'tty' and not 'cua'. Well, let's just
not filter serial ports for USB for now.
PR: ports/248227
Submitted by: greg@unrelenting.technology
The project has resumed since a few months by releasing one release every month (no more beta or pre release as between 2016 and 2020).
- cad/meshlab: Resurrect, update to 2020.05 and take maintainer'ship
- devel/vcglib: Update to @f7871ee [1]
[1] There is not new release, but Meshlab require this version
PR: 217821
Reported by: pi
Approved by: fernape, tcberner (mentor)
Differential Revision: https://reviews.freebsd.org/D24686
OpenEXR and ILMBASE updates to 2.5.2 rename the full shlib version
names from .0.0 to .0.1 => to be on the safe side, bump PORTREVISIONs.
Approved by: portmgr (blanket)
MFH: 2020Q3 (PORTREVISION bump after OpenEXR/ilmbase update)
Security: 714e6c35-c75b-11ea-aa29-d74973d1f9f3
Use newer GCC:
dstring.o: In function `copy':
dstring.c:(.opd+0x240): multiple definition of `copy'
spice2poly/dlmain.o:dlmain.c:(.opd+0x6f0): first defined here
MFH: 2020Q3 (fix build blanket)
Chase the devel/libffi update
Bump portrevision of all dependent ports to chace shard library version bump
in libffi.
Update LIB_DEPENDS lines where needed to not require a specific version of
libffi.so.
PR: 247028 (for tracking)
- While I'm here, add NO_ARCH
- Bump PORTREVISION of dependent ports for dependency change
(devel/py-pyvisa-py and finance/odoo will be bumped in next commit, py-pyusb renaming)
PR: 246546
Submitted by: sunpoet (myself)
Approved by: maintainer (timeout, 40 days)
/usr/local/bin/ld: src/twmc/CMakeFiles/TimberWolfMC.dir/readnets.c.o:(.bss+0x5e40): multiple definition of `yyss'; src/twmc/CMakeFiles/TimberWolfMC.dir/readcells.c.o:(.bss+0x5e40): first defined here
/usr/local/bin/ld: src/twmc/CMakeFiles/TimberWolfMC.dir/readnets.c.o:(.bss+0x6240): multiple definition of `yyssp'; src/twmc/CMakeFiles/TimberWolfMC.dir/readcells.c.o:(.bss+0x6240): first defined here
/usr/local/bin/ld: src/twmc/CMakeFiles/TimberWolfMC.dir/readnets.c.o:(.bss+0x4ea0): multiple definition of `yyvs'; src/twmc/CMakeFiles/TimberWolfMC.dir/readcells.c.o:(.bss+0x4ea0): first defined here
/usr/local/bin/ld: src/twmc/CMakeFiles/TimberWolfMC.dir/readnets.c.o:(.bss+0x6238): multiple definition of `yyvsp'; src/twmc/CMakeFiles/TimberWolfMC.dir/readcells.c.o:(.bss+0x6238): first defined here
http://package18.nyi.freebsd.org/data/121amd64-default-PR244494/2020-06-10_18h04m44s/logs/errors/graywolf-0.1.6.3_1.log
PR: 246700
In file included from /wrkdirs/usr/ports/cad/netgen/work/netgen-6.2.2004/libsrc/core/utils.hpp:14:
In file included from /usr/lib/clang/10.0.1/include/x86intrin.h:13:
/usr/lib/clang/10.0.1/include/ia32intrin.h:188:10: error: use of undeclared identifier '__builtin_ia32_readeflags_u32'
Approved by: portmgr (tier-2 blanket)
rcc automatically compresses resources above a certain size, but as
we found out the hard way, loading compressed resources from python
fails and may even coredump. As a workaround, force-disable compression
on resources in python code (c++ code seems to be fine compressed or not),
which re-enables several workbenches (most notably: Draft) at the cost
of a few megabytes of disk space.
Reported by: jeff@
MFH: 2020Q2
