← Back to team overview

kicad-developers team mailing list archive

Re: SIGTRAP crash with ngspice


I opened a bug here: https://bugzilla.redhat.com/show_bug.cgi?id=1666505

I've made the necessary changes in "fedora-packaging" but right now I'm getting an unreleated build failure:

[ 18%] Building CXX object common/CMakeFiles/pcbcommon.dir/__/pcbnew/io_mgr.cpp.o
cd /builddir/build/BUILD/kicad-r14777-54334233/common && /usr/bin/c++  -DGLM_FORCE_CTOR_INIT -DHAVE_STDINT_H -DKICAD_SCRIPTING -DKICAD_SCRIPTING_ACTION_MENU -DKICAD_SCRIPTING_MODULES -DKICAD_SPICE -DKICAD_USE_OCE -DWXUSINGDLL -DWX_COMPATIBILITY -D_FILE_OFFSET_BITS=64 -D__WXGTK__ -DPCBNEW -I/builddir/build/BUILD/kicad-r14777-54334233/include -I/builddir/build/BUILD/kicad-r14777-54334233/common/. -I/builddir/build/BUILD/kicad-r14777-54334233/common/./dialogs -I/builddir/build/BUILD/kicad-r14777-54334233/common/./widgets -I/builddir/build/BUILD/kicad-r14777-54334233/common/./dialog_about -I/usr/include/cairo -I/usr/include/pixman-1 -I/builddir/build/BUILD/kicad-r14777-54334233/common/../3d-viewer -I/builddir/build/BUILD/kicad-r14777-54334233/common/../pcbnew -I/builddir/build/BUILD/kicad-r14777-54334233/common/../polygon -I/builddir/build/BUILD/kicad-r14777-54334233 -I/usr/include/python2.7 -I/builddir/build/BUILD/kicad-r14777-54334233/scripting -I/usr/lib64/oce-0.18/../../include/oce -I/builddir/build/BUILD/kicad-r14777-54334233/common/../include/legacy_wx -isystem /usr/lib64/wx/include/gtk2-unicode-3.0 -isystem /usr/include/wx-3.0  -Wall -O2 -g -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fexceptions -fstack-protector-strong -grecord-gcc-switches -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -specs=/usr/lib/rpm/redhat/redhat-annobin-cc1 -m64 -mtune=generic -fasynchronous-unwind-tables -fstack-clash-protection -fcf-protection -Wsuggest-override -Werror=vla -Wno-unused-local-typedefs -Wno-strict-aliasing -pthread -Wshadow -g3 -ggdb3 -DDEBUG -Wno-deprecated-declarations -fPIC -fvisibility=hidden -fvisibility-inlines-hidden   -std=gnu++11 -o CMakeFiles/pcbcommon.dir/__/pcbnew/io_mgr.cpp.o -c /builddir/build/BUILD/kicad-r14777-54334233/pcbnew/io_mgr.cpp
BUILDSTDERR: /builddir/build/BUILD/kicad-r14777-54334233/pcbnew/io_mgr.cpp: In static member function 'static IO_MGR::PCB_FILE_T IO_MGR::GuessPluginTypeFromLibPath(const wxString&)':
BUILDSTDERR: /builddir/build/BUILD/kicad-r14777-54334233/pcbnew/io_mgr.cpp:164:19: error: 'GITHUB' was not declared in this scope
BUILDSTDERR:              ret = GITHUB;
BUILDSTDERR:                    ^~~~~~
BUILDSTDERR: /builddir/build/BUILD/kicad-r14777-54334233/pcbnew/io_mgr.cpp:164:19: note: suggested alternative: 'SIGHUP'
BUILDSTDERR:              ret = GITHUB;
BUILDSTDERR:                    ^~~~~~
BUILDSTDERR:                    SIGHUP
BUILDSTDERR: /builddir/build/BUILD/kicad-r14777-54334233/pcbnew/io_mgr.cpp: At global scope:
BUILDSTDERR: /builddir/build/BUILD/kicad-r14777-54334233/pcbnew/io_mgr.cpp:208:62: error: 'GITHUB' is not a member of 'IO_MGR'
BUILDSTDERR:  static IO_MGR::REGISTER_PLUGIN registerGithubPlugin( IO_MGR::GITHUB, wxT("Github"), []() -> PLUGIN* { return new GITHUB_PLUGIN; } );
BUILDSTDERR:                                                               ^~~~~~
BUILDSTDERR: make[2]: *** [common/CMakeFiles/pcbcommon.dir/build.make:556: common/CMakeFiles/pcbcommon.dir/__/pcbnew/io_mgr.cpp.o] Error 1

On 1/15/19 3:26 PM, Nick Østergaard wrote:
> Yes, I am fine with using it in the copr builds for now. Do we need to submit a PR for the libngspice package to let the packager pick it up more easy? I have not seen any activity from him.
> On Tue, 15 Jan 2019 at 20:58, Steven A. Falco <stevenfalco@xxxxxxxxx <mailto:stevenfalco@xxxxxxxxx>> wrote:
>     On 1/15/19 9:21 AM, Wayne Stambaugh wrote:
>     > Hi Steve,
>     >
>     > On 1/15/2019 8:51 AM, Steven A. Falco wrote:
>     >> I'll look at that.  The worst burst of backspaces is about 7 characters long, so I could accumulate 14 chars or so before making a decision.  In other words, run a circular buffer, and when I see the first non-backspace after a string of backspaces, then process the buffer.
>     >>
>     >> But I'm starting to think that the better approach is to drop this patch from the official tree, and just put my original patch into Fedora-only, as a temporary patch, to be removed when the library issue is corrected.
>     >
>     > This may be the way to go as this is only temporary until the ng-spice
>     > library is fixed.  I'm assuming this issue is specific to Fedora.  If
>     > not, we can re-evaluate it at the time that it is broken on another
>     > platform.
>     I committed my original patch into the Fedora build system.  A new build will appear in rawhide in a day or two, and it will appear in Fedora 29 after a week or two, once the karma process runs its course.  Thus Fedora users can start enjoying ngspice soon.  I'll keep track of the ngspice library, and if a fix appears there, then I'll remove my patch.
>     I did take a stab at a new, efficient patch which addresses Seth's concern - but after studying the dynamics of the message passing, I don't think it really matters.  Messages always seem to break on carriage-return boundaries, so either patch will work without any loss of data.
>     I attached the new patch here so it won't get lost.  It behaves about the same as the earlier one, but it does print some extra '%' characters that are left over after the backspaces are processed.  Thus, I actually prefer the original patch, because it produces slightly cleaner output.
>     Nick - Do you want to put either version into the Copr builds?
>             Steve
>     P.S. - This time, I tried to follow the coding style, as per Wayne's email. :-)

Follow ups