Activity for Stefan Brüns
-
Stefan Brüns posted a comment on ticket #622
This now (ngspice-42) apparently also affects x86_64. Output from kicad-7.0.10, on aarch64, ngspice-40: [ 1917s] + /usr/bin/ctest --output-on-failure --force-new-ctest-process -j8 --tests-regex qa_eeschema [ 1917s] Test project /home/abuild/rpmbuild/BUILD/kicad-7.0.10/build [ 1917s] Start 4: qa_eeschema [ 1923s] 1/1 Test #4: qa_eeschema ......................***Failed 6.32 sec [ 1923s] Running 104 test cases... [ 1923s] /home/abuild/rpmbuild/BUILD/kicad-7.0.10/qa/unittests/eeschema/./test_netlist_exporter_spice.h(242):...
-
highgui: Fix unresolved OpenGL functions for Qt backend
-
Duplicate of bug #8. https://sourceforge.net/p/ebook-tools/bugs/8/
-
Duplicate of bug #8.
-
The fixes are only syntactically correct, but not semantically. For the _cmp_ functions, 0 means "matching", but an empty id does not match "any" id.
-
The attached patch forces ghostscript to use the same Creation/ModDate as used in the PostScript preamble. The PDF file is still not fully deterministic as ghostscript uses the current time in some other places, but suffices to create a file which can be included in a TeX file without making the final output unreproducible. Notably the PDF trailer ID and DocumentUUID are not part of the final PDF produced by e.g. pdflatex.
-
Exported PDF is non-reproducible
-
libxfig.a contains src/main.c, and the test binaries are linked to libxfig, thus there are two definitions for "main". It would probably be better to leave src/main.c out of libxfig to avoid any ambiguities. You are relying on specific linker behavior, and you force it to not complain about multiple definitions: '-Wl,--allow-multiple-definitions'.
-
Btw, creating a fully emulated aarch64 machine on a x86_64 machine is fairly trivial with virt-manager. It shouldn't take more than 5 minutes to set one up (putting aside any download and installation time running in the background). Performance on a current x86_64 host is fairly decent, similar to an RPi3, so sufficient to build ngspice and simulate moderately sized circuits.
-
So, the op determination fails on aarch64 due to problems with numerical stability? Is this a problem of the circuit (no error margin when there should be some), or of the calculation (e.g. catastrophic error propagation)? I have seen differences between x87 FP, x86_64, PPC and AArch64 math over the years, so this is not totally unexpected, though most of the times adjusting error margins was sufficient. I am not specifically interested in making this circuit work on all architectures, more on deterministic...
-
Parser warning for voltage source
-
Error occurs with both ngspice-38 and ngspice-39.
-
KiCad device models: https://gitlab.com/kicad/code/kicad/-/tree/master/qa/data/eeschema Failing circuit netlist: i13162 KiCad schematic .include "./TL072-dual.lib" .include "./VDMOS_models.lib" .save all .probe alli .ic v(Tj1)={envtemp} v(Tj2)={envtemp} .temp {envtemp} .param envtemp=25 .tran 200u 500m .option RELTOL=.01 ABSTOL=1N VNTOL=10u .control set controlswait if $?sharedmode rusage else run rusage settype temperature tj1 tj2 tcase1 tcase2 plot tj1 tj2 tcase1 tcase2 plot in out xlimit 5.2 5.3...
-
Incorrect simulation results on aarch64
-
Missing libhdf5_hl when linking vigra plugin
-
spinit contains arch dependent paths, but is installed as noarch DATA
-
RUNPATH for robot clone libraries incorrect
-
This should be made conditional, based on HPDF_VERSION_MAJOR/MINOR, available from hpdf_version.h.
-
ngspice-37 Documentation
-
Fix signedness for ARM Neon mask type
-
libOpenGL does not export glActiveTextureARB, but only glActiveTexture. Both are different names for the same function (This is not true for all ARB extensions, but for glActiveTexture it is). glActiveTexture is a OpenGL 1.2 feature, so you can savely assume all implementations have it. Just replace glActiveTextureARB in tgfclient/gui.cpp with glActiveTexture.
-
FindOpenEXR.cmake incompatible with current OpenEXR
-
Proposed patch
-
Linking pfsglview fails with CMake >= 3.22
-
The following works, and is IMHO the correct way to specify the dependency:
-
https://sourceforge.net/p/pfstools/git/ci/master/tree/src/pfsview/pfsview_widget.cpp#l492 int p = binarySearchPixels( (*R)(index), lutPixFloor, lutSize ); pixel = lutPixel[p]; if( infNaNTreatment == INFNAN_MARK_AS_RED && (p == 0 || p == LUTSIZE+1)) As p converges to lutSize - 1 for large values, and lutSize = 257 * 2 + 1;, this is also an OOB access. So apparently float lutPixFloor[257 * 2 + 1] is correct. Also, p == LUTSIZE + 1 here (L494) should probably read p == lutSize - 1.
-
Out-of-bounds access in pfsview_widget
-
qhull: Use reentrant libqhull_r (bug #60016).
-
ngspice-35
-
Diode model name is parsed as numeric value in LT compat mode
-
Please provide a point release with all fixes from GIT
-
The following patch fixes the issue:
-
Compilation fails with GCC11/std=c++17, duplicate "clamp" definition
-
https://github.com/qhull/qhull/blob/master/CMakeLists.txt#L722 So something like ¨find_package(Qhull)", "target_link_libraries(... Qhull:qhull_r ....)" should be sufficient.
-
Use reentrant libqhull_r instead of deprecated libqhull
-
Note openSUSE 15.1 has been EOLed for quite some time. Leap 15.2 and Leap 15.3 (currently in late Beta) both contain the latest released version of PLPlot, 5.15.0.
-
Cleanup use of M_PI/PI defines
-
Remove two unused variables
-
Remove duplicated includes from occgeom.hpp
-
Remove occconstruction.cpp from list of library sources
-
Modernize code, replace Handle_<X> with Handle(X)
-
Fix GetTimeCounter for Aarch64 variants
-
cmpp man page is still installed
-
config.h gets installed by make install
-
https://sourceforge.net/p/openocd/tickets/286/#a5f3 We are going to deliver OpenOCD v0.11.0 soon, and the we should stop merging new features in ~10 days.
-
The ST-Link v3 is USB-HS (480MBit/s), so bandwidth should be no concern.
-
See also https://sourceforge.net/p/forge/feature-requests/614/ - no progress in 2.5 years.
-
Github supports this, Gitlab supports this ... SourceForge is making itself obsolete.
-
Fix detection of 32bit ARM builds on Aarch64
-
Explicitly default operator= for Vec<T, n>
-
Handle absolute OPENCV_INCLUDE_INSTALL_PATH correctly
-
See https://cmake.org/cmake/help/v3.15/policy/CMP0093.html
-
Boost 1.71.0 cmake config module implicitly removes PROJECT_BINARY_DIR from include_directories
-
Port ffmvforce to SDL2 (patch included)
-
epub_tit_next crashes on navPoint without navLabel
-
I think the correct place to specify the dependency is in the "DEPENDS" of the custom command. Otherwise, make may create e.g. pfsread.oct and libpfs in parallel. You only specify libpfs as a dependency of the .mkoct target.
-
Can the title be changed to SimGear fails to build with OpenSceneGraph >= 3.6.0 ? Related: https://sourceforge.net/p/flightgear/codetickets/1950/ - SG fails with OSG 3.5, fixed ...
-
Can the title be changed to SimGear fails to build with OpenSceneGraph >= 3.6.0 ? Related: https://sourceforge.net/p/flightgear/codetickets/2030/ - SG fails with OSG 3.5, fixed ...
-
For the building of software, not the library packages are relevant (i.e. anything you have listed above), but the accompanying *-devel packages, i.e. something like qt5-devel. Exact package names varies depending on distribution, also qt5 typically come with several -devel package for the independent modules. That said, the above error is not about a general incompatibility between FG and Qt5 (which definitely compiles and works fine), but specifying the wrong compiler flags (CFLAGS/CXXFLAGS) when...
-
Convert last README with DOS CRLF line endings
-
Avoid multiple explicit instantiations definition of templates
-
Regarding metadata: Comments: 79 occurences of "Created with GIMP", 4 occurences "Created with The GIMP", 11 occurences of "OrthoVista". Color profiles: yes, ICC profiles are essential information, unless, "look ma, my Photoshop can embed color profiles, lets embed a generic Adobe Profile".
-
can you please specify on which image you see a difference, and which options you have used? optipng does only lossless compression - if something different happens, that is a bug and should be reported.
-
I also added zopflipng to my tests. zopflipng does an aggressive optimization on the huffman tables, and is able to do another reduction of 5...15%.
-
The default settings of optipng do a completely lossless recompression, i.e. it uses a paletted format when only 256 colors or less are used, selects the best PNG filter, ... The decompressed RGB data is bit-for-bit identical. Optionally, optipng also allows to remove additional chunks, e.g. "Comment:Created with GIMP" or modify-date/create-date. Currently, the existing textures use RGB and RGBA truecolor and paletted images with varying depths, so I think all the options are covered and incompatibilities...
-
Recompress Textures with e.g. optipng
-
It provides an issue tracker and a git repository. Both are significantly better than the sourceforge implementations: it does not change formatting of messages at will it allows tagging of issues it nicely links between pull requests and issues it allows to download commits as patch files (required for backporting e.g. compiler warning fixes). * it makes creating pull requests much easier There is nothing wrong with leaving e.g. the ML at SF, but move the code to github.
-
Randomly selected graphic used for "Warning" admonition
-
The "shift" above was still wrong, in the attached patch it is correct (i.e. shift = 8 - bpc).
-
Bad static casts in code
1