Date: Tue, 7 Apr 2026 08:28:16 +0200 From: Gleb Popov <arrowd@freebsd.org> To: Vladimir Druzenko <vvd@freebsd.org> Cc: Charlie Li <vishwin@freebsd.org>, ports-committers <ports-committers@freebsd.org>, dev-commits-ports-all@freebsd.org, dev-commits-ports-main@freebsd.org Subject: Re: git: ef9180aa8469 - main - science/py-scipy: update to 1.17.1 Message-ID: <CALH631kW-KxA2t6cTu8JY7ztV8cR1b2qYtDhcnuJwMWup0cGHg@mail.gmail.com> In-Reply-To: <8648dad5-11cc-4a79-83e2-47edfb5b99e1@freebsd.org> References: <69d03cf2.430ed.2d4d8408@gitrepo.freebsd.org> <bb7cb6b8-4d8b-423e-88dc-f67629fc0a25@freebsd.org> <37778713-0656-4edf-83d5-0b7a84bceb8e@freebsd.org> <cd2d82b3-8626-4aa1-822a-617b5a5dc410@freebsd.org> <3d372c89-9f77-4033-80e0-6cec0781d568@freebsd.org> <8648dad5-11cc-4a79-83e2-47edfb5b99e1@freebsd.org>
index | next in thread | previous in thread | raw e-mail
[-- Attachment #1 --] On Tue, Apr 7, 2026, 02:54 Vladimir Druzenko <vvd@freebsd.org> wrote: > 05.04.2026 00:54, Charlie Li пишет: > > Vladimir Druzenko wrote: > >> Thanks for the explanation. After updating of math/py-numpy building > >> ports that depended on it failed with error. And it was not at all > >> obvious what to do. > >> > >> But build of the science/py-scipy fails: > >> [846/1238] Compiling C++ object scipy/fft/_pocketfft/ > >> pypocketfft.cpython-311.so.p/pypocketfft.cxx.o > >> FAILED: [code=1] scipy/fft/_pocketfft/pypocketfft.cpython-311.so.p/ > >> pypocketfft.cxx.o > >> c++ -Iscipy/fft/_pocketfft/pypocketfft.cpython-311.so.p -Iscipy/fft/ > >> _pocketfft -I../scipy/fft/_pocketfft -I../scipy/_lib/pocketfft -I/ > >> include -I/usr/local/include/python3.11 -fvisibility=hidden - > >> fvisibility-inlines-hidden -fdiagnostics-color=always -DNDEBUG - > >> D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -std=c++17 -O3 -O2 -pipe - > >> march=core2 -fstack-protector-strong -fno-strict-aliasing -fPIC > >> -pthread -DPOCKETFFT_PTHREADS -DPOCKETFFT_CACHE_SIZE=16 -MD -MQ > >> scipy/fft/ _pocketfft/pypocketfft.cpython-311.so.p/pypocketfft.cxx.o > >> -MF scipy/fft/ > >> _pocketfft/pypocketfft.cpython-311.so.p/pypocketfft.cxx.o.d -o scipy/ > >> fft/_pocketfft/pypocketfft.cpython-311.so.p/pypocketfft.cxx.o -c ../ > >> scipy/fft/_pocketfft/pypocketfft.cxx > >> ../scipy/fft/_pocketfft/pypocketfft.cxx:16:10: fatal error: > >> 'pybind11/ pybind11.h' file not found > >> 16 | #include <pybind11/pybind11.h> > >> | ^~~~~~~~~~~~~~~~~~~~~ > >> 1 error generated. > >> > >> USES+=localbase fixes the build. > >> > > Do you have cmake in the environment used for this build run? If so, > > remove it and try again. meson will prefer to use cmake for dependency > > detection over pkg-config and other methods if cmake is present. > > > > In this specific case, devel/py-pybind11 is present exclusively in the > > Python package hierarchy, specifically site-packages, and not in the > > system. meson needs to fall back to methods that are not cmake for the > > compiler flags to set properly. > > > > This part is unspoken, but bears speaking and documenting somewhere > > better than bugzilla and mailing lists: building Python packages on a > > live environment, ie without environment separation like virtual > > environments or jails (poudriere included) is not supported, even if > > the ports framework makes it technically possible to try. > > > You just wrote that one of the main advantages of FreeBSD over other > operating systems "is not supported". > Building in an non-isolated environment is certainly not an advantage and even does not have anything to do with the OS. It makes me very sad to see how FreeBSD is being turned into yet another > Linux distribution, but with the worst compatibility with other Linux > distributions… > > -- > Best regards, > Vladimir Druzenko > > [-- Attachment #2 --] <div dir="auto"><div><br><br><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">On Tue, Apr 7, 2026, 02:54 Vladimir Druzenko <<a href="mailto:vvd@freebsd.org">vvd@freebsd.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">05.04.2026 00:54, Charlie Li пишет:<br> > Vladimir Druzenko wrote:<br> >> Thanks for the explanation. After updating of math/py-numpy building <br> >> ports that depended on it failed with error. And it was not at all <br> >> obvious what to do.<br> >><br> >> But build of the science/py-scipy fails:<br> >> [846/1238] Compiling C++ object scipy/fft/_pocketfft/ <br> >> pypocketfft.cpython-311.so.p/pypocketfft.cxx.o<br> >> FAILED: [code=1] scipy/fft/_pocketfft/pypocketfft.cpython-311.so.p/ <br> >> pypocketfft.cxx.o<br> >> c++ -Iscipy/fft/_pocketfft/pypocketfft.cpython-311.so.p -Iscipy/fft/ <br> >> _pocketfft -I../scipy/fft/_pocketfft -I../scipy/_lib/pocketfft -I/ <br> >> include -I/usr/local/include/python3.11 -fvisibility=hidden - <br> >> fvisibility-inlines-hidden -fdiagnostics-color=always -DNDEBUG - <br> >> D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -std=c++17 -O3 -O2 -pipe - <br> >> march=core2 -fstack-protector-strong -fno-strict-aliasing -fPIC <br> >> -pthread -DPOCKETFFT_PTHREADS -DPOCKETFFT_CACHE_SIZE=16 -MD -MQ <br> >> scipy/fft/ _pocketfft/pypocketfft.cpython-311.so.p/pypocketfft.cxx.o <br> >> -MF scipy/fft/ <br> >> _pocketfft/pypocketfft.cpython-311.so.p/pypocketfft.cxx.o.d -o scipy/ <br> >> fft/_pocketfft/pypocketfft.cpython-311.so.p/pypocketfft.cxx.o -c ../ <br> >> scipy/fft/_pocketfft/pypocketfft.cxx<br> >> ../scipy/fft/_pocketfft/pypocketfft.cxx:16:10: fatal error: <br> >> 'pybind11/ pybind11.h' file not found<br> >> 16 | #include <pybind11/pybind11.h><br> >> | ^~~~~~~~~~~~~~~~~~~~~<br> >> 1 error generated.<br> >><br> >> USES+=localbase fixes the build.<br> >><br> > Do you have cmake in the environment used for this build run? If so, <br> > remove it and try again. meson will prefer to use cmake for dependency <br> > detection over pkg-config and other methods if cmake is present.<br> ><br> > In this specific case, devel/py-pybind11 is present exclusively in the <br> > Python package hierarchy, specifically site-packages, and not in the <br> > system. meson needs to fall back to methods that are not cmake for the <br> > compiler flags to set properly.<br> ><br> > This part is unspoken, but bears speaking and documenting somewhere <br> > better than bugzilla and mailing lists: building Python packages on a <br> > live environment, ie without environment separation like virtual <br> > environments or jails (poudriere included) is not supported, even if <br> > the ports framework makes it technically possible to try.<br> ><br> You just wrote that one of the main advantages of FreeBSD over other <br> operating systems "is not supported".<br></blockquote></div></div><div dir="auto"><br></div><div dir="auto">Building in an non-isolated environment is certainly not an advantage and even does not have anything to do with the OS.</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote gmail_quote_container"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> It makes me very sad to see how FreeBSD is being turned into yet another <br> Linux distribution, but with the worst compatibility with other Linux <br> distributions…<br> <br> -- <br> Best regards,<br> Vladimir Druzenko<br> <br> </blockquote></div></div></div>home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CALH631kW-KxA2t6cTu8JY7ztV8cR1b2qYtDhcnuJwMWup0cGHg>