While there are certainly build environments that do work to build
gap_packages, it is clearly not robust. I think it would be a mistake to
ignore this, because it is guaranteed to produce lots of support emails.
The issue here is with the file TextFlow.hpp. When libsemigroups is
installed, that file ends up with the following path
$SAGE_LOCAL/include/libsemigroups/textflowcpp/TextFlow.hpp
I also see the file in
$SAGE_LOCAL/lib/gap/pkg/semigroups/libsemigroups/extern/textflowcpp/TextFlow.hpp
I think the first one is the sensible one to use. We can count on
libsemigroups being installed before gap_packages because gap_packages
lists it as a dependency.
If you look at the g++ command which is failing, it specifies the
directories to be searched for headers with:
-I./bin/include
-I./bin/include/libsemigroups
-Igapbind14/include/
-Ilibsemigroups/extern/HPCombi/include
-Ilibsemigroups/extern/HPCombi/include/fallback
-Ilibsemigroups/extern/fmt-8.0.1/include
-Ilibsemigroups/include
-I/usr/local/sage-10/local/include/gap
-I/usr/local/sage-10/local/include
That list does not include $SAGE_LOCAL/include/libsemigroups/textflowcpp
(I am guessing that in this case $SAGE_LOCAL == /usr/local/sage-10)
I think it might fix this problem to add the following line at the top of
build/pkgs/gap_packages/spkg-install.in:
export CFLAGS="${CFLAGS} -I${SAGE_LOCAL}/include/libsemigroups/textflowcpp"
Emmanuel, would you try that?
(Of course it might only reveal that there is another -I option needed, but
that would be progress.)
- Marc
On Tuesday, October 29, 2024 at 8:35:28 AM UTC-5 [email protected]
wrote:
> Le mardi 29 octobre 2024 à 10:36:17 UTC+1, Dima Pasechnik a écrit :
>
> On Tue, Oct 29, 2024 at 8:59 AM Emmanuel Charpentier
> <[email protected]> wrote:
> >
> > The updated PR (re-fetched at Tue Oct 29 09:22:19 CET 2024) does not
> solve the problem. Without make dist-clean, I get the same problem as
> before :
>
> It certainly does the job after `make distclean`,
>
> Nope.
>
> *After make distclean*, ./bootstrap,,configure $(./config.status --config,
> make -j8 build still triggers the same problem :
> [gap_packages-4.13.1] [spkg-install] g++ -std=gnu++11 -pthread -g -O2
> -fPIC -MQ gen/src/bipart.o -MMD -MP -MF gen/src/bipart.d -I./bin/include
> -I./bin/include/libsemigroups -std=gnu++14 -O3 -mavx
> -flax-vector-conversions -Igapbind14/include/ -DHPCOMBI_CONSTEXPR_FUN_ARGS
> -Ilibsemigroups/extern/HPCombi/include
> -Ilibsemigroups/extern/HPCombi/include/fallback
> -Ilibsemigroups/extern/fmt-8.0.1/include -Ilibsemigroups/include
> -DFMT_HEADER_ONLY -DNDEBUG -g -O2 -o gen/src/bipart.o
> -I/usr/local/sage-10/local/include/gap -I/usr/local/sage-10/local/include
> -DUSE_GASMAN=1 -c src/bipart.cpp [gap_packages-4.13.1] [spkg-install] In
> file included from src/bipart.cpp:38: [gap_packages-4.13.1] [spkg-install]
> libsemigroups/include/libsemigroups/report.hpp:42:10: fatal error:
> textflowcpp/TextFlow.hpp: No such file or directory [gap_packages-4.13.1]
> [spkg-install] 42 | #include "textflowcpp/TextFlow.hpp"
> [gap_packages-4.13.1] [spkg-install] | ^~~~~~~~~~~~~~~~~~~~~~~~~~
> [gap_packages-4.13.1] [spkg-install] compilation terminated.
>
> HTH,
>
> and we have hard
> time figuring out how exactly the old state
> (the presence of traces of Eigen headers somewhere deep in the guts of
> the build system) is cached.
>
> I'd actually propose to add Eigen as an optional package, but
> meanwhile we need a PR to unblock the built
> of gap_packages.
>
>
> >
> > [gap_packages-4.13.1] [spkg-install] g++ -std=gnu++11 -pthread -g -O2
> -fPIC -MQ gen/src/cong.o -MMD -MP -MF gen/src/cong.d -std=gnu++14 -O3 -mavx
> -flax-vector-conversions -Igapbind14/include/ -DHPCOMBI_CONSTEXPR_FUN_ARGS
> -DFMT_HEADER_ONLY -DNDEBUG -g -O2 -o gen/src/cong.o
> -I/usr/local/sage-10/local/include/gap -I/usr/local/sage-10/local/include
> -DUSE_GASMAN=1 -c src/cong.cpp [gap_packages-4.13.1] [spkg-install] In file
> included from
> /usr/local/sage-10/local/include/libsemigroups/knuth-bendix.hpp:34,
> [gap_packages-4.13.1] [spkg-install] from
> /usr/local/sage-10/local/include/libsemigroups/cong.hpp:30,
> [gap_packages-4.13.1] [spkg-install] from src/to_cpp.hpp:54,
> [gap_packages-4.13.1] [spkg-install] from src/cong.cpp:29:
> [gap_packages-4.13.1] [spkg-install]
> /usr/local/sage-10/local/include/libsemigroups/digraph.hpp:58:10: fatal
> error: Eigen/Core: No such file or directory [gap_packages-4.13.1]
> [spkg-install] 58 | #include <Eigen/Core> [gap_packages-4.13.1]
> [spkg-install] | ^~~~~~~~~~~~ [gap_packages-4.13.1] [spkg-install]
> compilation terminated.
> >
> > HTH,
> >
> > Le lundi 28 octobre 2024 à 18:15:37 UTC+1, [email protected] a écrit :
> >>
> >> I agree that this in not fixed, in spite of my positive review of the
> PR.
> >>
> >> I did not get that error, but I see by looking at the timestamps that
> my sage/local/include/libsemigroups/Eigen directory was left over from the
> install of the previous version.
> >>
> >> Michael Orlitzky points out that the libsemigroups package is
> configured with --disable-eigen. But apparently that option does not work
> correctly.
> >>
> >> - Marc
> >>
> >> On Monday, October 28, 2024 at 11:59:18 AM UTC-5 [email protected]
> wrote:
> >>>
> >>> The relevant error is different :
> >>>
> >>> ```
> >>> [gap_packages-4.13.1] [spkg-install] g++ -std=gnu++11 -pthread -g -O2
> -fPIC -MQ gen/src/cong.o -MMD -MP -MF gen/src/cong.d -std=gnu++14 -O3 -mavx
> -flax-vector-conversions -Igapbind14/include/ -DHPCOMBI_CONSTEXPR_FUN_ARGS
> -DFMT_HEADER_ONLY -DNDEBUG -g -O2 -o gen/src/cong.o
> -I/usr/local/sage-10/local/include/gap -I/usr/local/sage-10/local/include
> -DUSE_GASMAN=1 -c src/cong.cpp
> >>> [gap_packages-4.13.1] [spkg-install] In file included from
> /usr/local/sage-10/local/include/libsemigroups/knuth-bendix.hpp:34,
> >>> [gap_packages-4.13.1] [spkg-install] from
> /usr/local/sage-10/local/include/libsemigroups/cong.hpp:30,
> >>> [gap_packages-4.13.1] [spkg-install] from src/to_cpp.hpp:54,
> >>> [gap_packages-4.13.1] [spkg-install] from src/cong.cpp:29:
> >>> [gap_packages-4.13.1] [spkg-install]
> /usr/local/sage-10/local/include/libsemigroups/digraph.hpp:58:10: fatal
> error: Eigen/Core: No such file or directory
> >>> [gap_packages-4.13.1] [spkg-install] 58 | #include <Eigen/Core>
> >>> [gap_packages-4.13.1] [spkg-install] | ^~~~~~~~~~~~
> >>> [gap_packages-4.13.1] [spkg-install] compilation terminated.
> >>> [gap_packages-4.13.1] [spkg-install] make[5]: ***
> [Makefile.gappkg:130: gen/src/cong.o] Error 1
> >>> ```
> >>>
> >>> HTH,
> >>>
> >>> Le lundi 28 octobre 2024 à 17:48:45 UTC+1, Emmanuel Charpentier a
> écrit :
> >>>>
> >>>> Le lundi 28 octobre 2024 à 16:58:55 UTC+1, Dima Pasechnik a écrit :
> >>>>
> >>>> The libsemigroup issue is fixed by
> https://github.com/sagemath/sage/pull/38875
> >>>> (the version bump was forgotten, basically)
> >>>>
> >>>> Please test and review
> >>>>
> >>>>
> >>>> Nope, make -j8 build failed. Logs follow
> >>>>
> >>>>
> >>>>
> >>>> On Mon, Oct 28, 2024 at 2:41 PM Emmanuel Charpentier
> >>>> <[email protected]> wrote:
> >>>> >
> >>>> > The promised logs.
> >>>> >
> >>>> > HTH,
> >>>> >
> >>>> > --
> >>>> > You received this message because you are subscribed to the Google
> Groups "sage-release" group.
> >>>> > To unsubscribe from this group and stop receiving emails from it,
> send an email to [email protected].
> >>>> > To view this discussion visit
> https://groups.google.com/d/msgid/sage-release/CAFG8T6UQUYiCts1jrxNcU1T_1scc9rHdjZuMOHrVQHCf0_pXEQ%40mail.gmail.com.
>
>
> >
> > --
> > You received this message because you are subscribed to the Google
> Groups "sage-release" group.
> > To unsubscribe from this group and stop receiving emails from it, send
> an email to [email protected].
> > To view this discussion visit
> https://groups.google.com/d/msgid/sage-release/74f36963-6cce-4e4e-be6f-407238897541n%40googlegroups.com.
>
>
>
>
>
--
You received this message because you are subscribed to the Google Groups
"sage-release" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion visit
https://groups.google.com/d/msgid/sage-release/73a8949c-55c9-4635-9972-745761042f1fn%40googlegroups.com.
