Hi, This might be an environment issue, but I suspect missing dependency. >>> Emerging (2 of 163) dev-ml/ocaml-ctypes-0.23.0::gentoo >>> Failed to emerge dev-ml/ocaml-ctypes-0.23.0, Log file: >>> '/var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/temp/build.log' >>> Jobs: 1 of 163 complete, 1 failed Load avg: 7.31, 3.42, 1.75 * Package: dev-ml/ocaml-ctypes-0.23.0:0/0.23.0 * Repository: gentoo * Maintainer: ml@gentoo.org * USE: abi_x86_64 amd64 elibc_glibc kernel_linux ocamlopt * FEATURES: network-sandbox preserve-libs sandbox userpriv usersandbox >>> Unpacking source... >>> Unpacking ocaml-ctypes-0.23.0.tar.gz to /var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/work >>> Source unpacked in /var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/work >>> Preparing source in /var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/work/ocaml-ctypes-0.23.0 ... >>> Source prepared. >>> Configuring source in /var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/work/ocaml-ctypes-0.23.0 ... >>> Source configured. >>> Compiling source in /var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/work/ocaml-ctypes-0.23.0 ... * dune build @install --display=short --profile release -j 6 ocamlc src/ctypes-top/.ctypes_top.objs/byte/ctypes_top.{cmi,cmo,cmt} ocamlopt src/ctypes-top/.ctypes_top.objs/native/ctypes_top.{cmx,o} ... discover src/ctypes-foreign/backend.sexp,src/ctypes-foreign/c_flags,src/ctypes-foreign/c_flags.sexp,src/ctypes-foreign/c_library_flags.sexp File "src/ctypes-foreign/dune", line 25, characters 33-47: 25 | (names dl_stubs ffi_call_stubs ffi_type_stubs foreign_threaded_stubs) ^^^^^^^^^^^^^^ x86_64-pc-linux-gnu-gcc src/ctypes-foreign/ffi_type_stubs.o (exit 1) ffi_type_stubs.c:14:10: fatal error: ffi.h: No such file or directory 14 | #include <ffi.h> | ^~~~~~~ compilation terminated. File "src/ctypes-foreign/dune", line 25, characters 18-32: 25 | (names dl_stubs ffi_call_stubs ffi_type_stubs foreign_threaded_stubs) ^^^^^^^^^^^^^^ x86_64-pc-linux-gnu-gcc src/ctypes-foreign/ffi_call_stubs.o (exit 1) ffi_call_stubs.c:22:10: fatal error: ffi.h: No such file or directory 22 | #include <ffi.h> | ^~~~~~~ compilation terminated. Full build.log will be attached momentarily. Reproducible: Always
Created attachment 913617 [details] build.log
I dropped all dev-ml packages and emerge -D ocaml-ctypes and it builds correctly The ffi.h is in libffi that is in DEPEND/RDEPEND So the ebuild seems good. If you want to help drop all dev-ml packages and then try: emerge -aDv ocaml-ctypes List all the packages it is emerging, so I can try with your version
Created attachment 914329 [details] emerge --info ocaml-ctypes (In reply to Tupone Alfredo from comment #2) > I dropped all dev-ml packages and emerge -D ocaml-ctypes and it builds > correctly > > The ffi.h is in libffi that is in DEPEND/RDEPEND * dev-libs/libffi Latest version available: 3.4.6 Latest version installed: 3.4.6 Size of files: 1,360 KiB Homepage: https://sourceware.org/libffi/ Description: Portable, high level programming interface to various calling conventions License: MIT And installed, but still I get: File "src/ctypes-foreign/dune", line 25, characters 33-47: 25 | (names dl_stubs ffi_call_stubs ffi_type_stubs foreign_threaded_stubs) ^^^^^^^^^^^^^^ x86_64-pc-linux-gnu-gcc src/ctypes-foreign/ffi_type_stubs.o (exit 1) ffi_type_stubs.c:14:10: fatal error: ffi.h: No such file or directory 14 | #include <ffi.h> | ^~~~~~~ compilation terminated. x86_64-pc-linux-gnu-gcc src/ctypes-foreign/foreign_threaded_stubs.o File "src/ctypes-foreign/dune", line 25, characters 18-32: 25 | (names dl_stubs ffi_call_stubs ffi_type_stubs foreign_threaded_stubs) ^^^^^^^^^^^^^^ x86_64-pc-linux-gnu-gcc src/ctypes-foreign/ffi_call_stubs.o (exit 1) ffi_call_stubs.c:22:10: fatal error: ffi.h: No such file or directory 22 | #include <ffi.h> | ^~~~~~~ > So the ebuild seems good. As mentioned, may be my environment. But I cannot see what. I've attached emerge --info > If you want to help drop all dev-ml packages and then try: > > emerge -aDv ocaml-ctypes Doesn't help. > List all the packages it is emerging, so I can try with your version [ebuild rR ] dev-ml/camlzip-1.11:1/1.11::gentoo USE="ocamlopt" 28 KiB [ebuild r U ] dev-ml/camlbz2-0.8.0-r1:0/0.8.0::gentoo [0.7.0:0/0.7.0::gentoo] USE="ocamlopt" 17 KiB [ebuild r U ] dev-ml/ocaml-ctypes-0.23.0:0/0.23.0::gentoo [0.20.1:0/0.20.1::gentoo] USE="ocamlopt -test" 0 KiB [ebuild rR ] dev-ml/parmap-1.2.5:0/1.2.5::gentoo USE="ocamlopt" 0 KiB [ebuild r U ] dev-ml/ocamlgraph-2.1.0:0/2.1.0::gentoo [2.0.0-r1:0/2.0.0::gentoo] USE="ocamlopt" 0 KiB [ebuild r U ] dev-ml/luv-0.5.14-r1:0/0.5.14::gentoo [0.5.12:0/0.5.12::gentoo] USE="ocamlopt -test" 0 KiB [ebuild rR ] dev-ml/ppx_here-0.16.0:0/0.16::gentoo USE="ocamlopt" 0 KiB [ebuild rR ] dev-ml/ppx_let-0.16.0:0/0.16::gentoo USE="ocamlopt" 0 KiB [ebuild rR ] dev-ml/ocaml-gettext-0.4.2-r1:0/0.4.2::gentoo USE="ocamlopt -test" 0 KiB [ebuild rR ] dev-ml/ocaml-gettext-stub-0.4.2-r2:0/0.4.2::gentoo USE="ocamlopt -test" 0 KiB [ebuild r U ] dev-ml/dose3-7.0.0-r1:0/7.0.0::gentoo [7.0.0:0/7.0.0::gentoo] USE="ocamlopt" 0 KiB [ebuild rR ] dev-ml/mccs-1.1.17:0/1.1.17::gentoo USE="ocamlopt -test" 0 KiB [ebuild rR ] dev-ml/opam-2.1.6:0/2.1.6::gentoo USE="ocamlopt" 0 KiB [ebuild rR ] dev-ml/opam-installer-2.1.6:0/2.1.6::gentoo USE="ocamlopt" 0 KiB [ebuild rR ] dev-ml/react-1.2.2-r1:0/1.2.2::gentoo USE="-test" 0 KiB [ebuild rR ] dev-ml/lwt-5.7.0:0/5.7.0::gentoo USE="ocamlopt" 0 KiB [ebuild rR ] dev-ml/ounit2-2.2.7-r1:0/2.2.7::gentoo USE="ocamlopt" 0 KiB [ebuild rR ] app-emulation/libguestfs-1.48.6-r1:0/1.48::gentoo USE="fuse gtk ocaml -doc -erlang -inspect-icons -introspection -libvirt -lua -perl -python -ruby (-selinux) -static-libs -systemtap -test" LUA_SINGLE_TARGET="lua5-1" PYTHON_SINGLE_TARGET="python3_12 -python3_10 -python3_11" 0 KiB [ebuild rR ] app-emulation/guestfs-tools-1.48.2:0/1.48::gentoo USE="ocaml -doc -perl -test" 0 KiB Failed in above mode. Any other package lists that may be helpful?
I don't see dev-ml/dune. So I don't think you removed all the dev-ml packages. If you want you can go to /var/db/pkg and then emerge -C dev-ml/* And the command to get the list of dev-ml packages: emerge -av dev-ml/ocaml-ctypes
plastiekpoot [13:48:01] ~ # emerge -C $(qlist -IC dev-ml/*) ... >>> Unmerging (1 of 46) dev-ml/base-0.16.3-r1... ... >>> Unmerging (46 of 46) dev-ml/xml-light-2.5... plastiekpoot [13:49:35] ~ # emerge -aDv ocaml-ctypes ... Dependency resolution took 4.92 s (backtrack: 0/20). [ebuild N ] dev-ml/dune-3.16.0:0/3.16.0::gentoo USE="-emacs" 0 KiB [ebuild N ] dev-ml/bigarray-compat-1.1.0::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/result-1.5:0/1.5::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/stdlib-shims-0.3.0::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/integers-0.7.0:0/0.7.0::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/csexp-1.5.2-r1:0/1.5.2::gentoo USE="ocamlopt -test" 0 KiB [ebuild N ] dev-ml/dune-configurator-3.16.0-r1:0/3.16.0::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/ocaml-ctypes-0.23.0:0/0.23.0::gentoo USE="ocamlopt -test" 0 KiB Total: 8 packages (8 new), Size of downloads: 0 KiB Would you like to merge these packages? [Yes/No] ... >>> Completed (7 of 8) dev-ml/dune-configurator-3.16.0-r1::gentoo >>> Emerging (8 of 8) dev-ml/ocaml-ctypes-0.23.0::gentoo ... and failure mode remains unaltered. Happy to try more suggestions/action further trouble shooting, just need to know which rabbit hole to start down.
I have built exactly with your version of packages with ocaml-4.14.2, and all goes smooth, It does not seems a missing dependency. I hope I could find a problem here so I can fix it, but no When you say "This might be an environment issue" what do you mean?
I mean it may be something weird on my laptop. Having said that, I just bumped into same on another machine where I'm not quite so adventurous with setting random stuff. arthur [09:50:18] ~ # emerge -C $(qlist -IC dev-ml/*) ... removed 7 packages. arthur [09:50:46] ~ # emerge -aDv ocaml-ctypes ... [ebuild N ] dev-ml/dune-3.16.0:0/3.16.0::gentoo USE="-emacs" 0 KiB [ebuild N ] dev-ml/bigarray-compat-1.1.0::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/result-1.5:0/1.5::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/stdlib-shims-0.3.0::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/integers-0.7.0:0/0.7.0::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/csexp-1.5.2-r1:0/1.5.2::gentoo USE="ocamlopt -test" 0 KiB [ebuild N ] dev-ml/dune-configurator-3.16.0-r1:0/3.16.0::gentoo USE="ocamlopt" 0 KiB [ebuild N ] dev-ml/ocaml-ctypes-0.23.0:0/0.23.0::gentoo USE="ocamlopt -test" 0 KiB ... File "src/ctypes-foreign/dune", line 25, characters 18-32: 25 | (names dl_stubs ffi_call_stubs ffi_type_stubs foreign_threaded_stubs) ^^^^^^^^^^^^^^ x86_64-pc-linux-gnu-gcc src/ctypes-foreign/ffi_call_stubs.o (exit 1) ffi_call_stubs.c:22:10: fatal error: ffi.h: No such file or directory 22 | #include <ffi.h> | ^~~~~~~ compilation terminated. File "src/ctypes-foreign/dune", line 25, characters 33-47: 25 | (names dl_stubs ffi_call_stubs ffi_type_stubs foreign_threaded_stubs) ^^^^^^^^^^^^^^ x86_64-pc-linux-gnu-gcc src/ctypes-foreign/ffi_type_stubs.o (exit 1) ffi_type_stubs.c:14:10: fatal error: ffi.h: No such file or directory 14 | #include <ffi.h> | ^~~~~~~ jkroon@plastiekpoot ~ $ find /usr/include/ -name ffi.h /usr/include/botan-3/botan/ffi.h /usr/include/botan-2/botan/ffi.h However, on arthur: arthur [10:04:39] ~ # find /usr/include/ -name ffi.h arthur [10:09:04] ~ # Both systems has libffi installed (3.4.6), but neither one has a ffi.h file installed in /usr/include, however: jkroon@plastiekpoot ~ $ equery files libffi | grep ffi[.]h /usr/lib64/libffi/include/ffi.h arthur [10:09:39] ~ # equery files libffi | grep ffi[.]h /usr/lib64/libffi/include/ffi.h So either we're missing some compile option here to *find* that location, or libffi isn't installing into the correct location (IMHO). Looking at the libffi ebuild, it explicitly passes --includedir="${EPREFIX}"/usr/$(get_libdir)/${PN}/include to econf. Regardless, pkg-config returns the appropriate cflags: jkroon@plastiekpoot ~ $ pkg-config --cflags libffi -I/usr/lib64/libffi/include And just for completeness, --libs also seems OK. jkroon@plastiekpoot ~ $ pkg-config --libs libffi -lffi jkroon@plastiekpoot ~ $ equery files libffi | grep libffi /usr/lib64/libffi /usr/lib64/libffi.so I'm somewhat stumped.
Try to add the following line to the ebuild, to show the real commands dune is running. Maybe you can get some hints: src_compile() { dune build @install --verbose --profile release -j 1 }
Created attachment 914496 [details] verbose build.log The colour output doesn't help, but there you have it, these are the two that's problematic: Running[113]: (cd _build/default/src/ctypes-foreign && /usr/bin/x86_64-pc-linux-gnu-gcc -O2 -fno-strict-aliasing -fwrapv -pthread -fPIC -O2 -march=native -pipe -D_FILE_OFFSET_BITS=64 -fdiagnostics-color=always -g -I /usr/lib64/ocaml -I /usr/lib64/ocaml/bigarray-compat -I /usr/lib64/ocaml/integers -I /usr/lib64/ocaml/stdlib-shims -I /usr/lib64/ocaml/threads -I ../ctypes -o ffi_call_stubs.o -c ffi_call_stubs.c) File "src/ctypes-foreign/dune", line 25, characters 18-32: 25 | (names dl_stubs ffi_call_stubs ffi_type_stubs foreign_threaded_stubs) ^^^^^^^^^^^^^^ Command [113] exited with code 1: $ (cd _build/default/src/ctypes-foreign && /usr/bin/x86_64-pc-linux-gnu-gcc -O2 -fno-strict-aliasing -fwrapv -pthread -fPIC -O2 -march=native -pipe -D_FILE_OFFSET_BITS=64 -fdiagnostics-color=always -g -I /usr/lib64/ocaml -I /usr/lib64/ocaml/bigarray-compat -I /usr/lib64/ocaml/integers -I /usr/lib64/ocaml/stdlib-shims -I /usr/lib64/ocaml/threads -I ../ctypes -o ffi_call_stubs.o -c ffi_call_stubs.c) ffi_call_stubs.c:22:10: fatal error: ffi.h: No such file or directory 22 | #include <ffi.h> | ^~~~~~~ compilation terminated. Running[114]: (cd _build/default/src/ctypes-foreign && /usr/bin/x86_64-pc-linux-gnu-gcc -O2 -fno-strict-aliasing -fwrapv -pthread -fPIC -O2 -march=native -pipe -D_FILE_OFFSET_BITS=64 -fdiagnostics-color=always -g -I /usr/lib64/ocaml -I /usr/lib64/ocaml/bigarray-compat -I /usr/lib64/ocaml/integers -I /usr/lib64/ocaml/stdlib-shims -I /usr/lib64/ocaml/threads -I ../ctypes -o ffi_type_stubs.o -c ffi_type_stubs.c) File "src/ctypes-foreign/dune", line 25, characters 33-47: 25 | (names dl_stubs ffi_call_stubs ffi_type_stubs foreign_threaded_stubs) ^^^^^^^^^^^^^^ Command [114] exited with code 1: $ (cd _build/default/src/ctypes-foreign && /usr/bin/x86_64-pc-linux-gnu-gcc -O2 -fno-strict-aliasing -fwrapv -pthread -fPIC -O2 -march=native -pipe -D_FILE_OFFSET_BITS=64 -fdiagnostics-color=always -g -I /usr/lib64/ocaml -I /usr/lib64/ocaml/bigarray-compat -I /usr/lib64/ocaml/integers -I /usr/lib64/ocaml/stdlib-shims -I /usr/lib64/ocaml/threads -I ../ctypes -o ffi_type_stubs.o -c ffi_type_stubs.c) ffi_type_stubs.c:14:10: fatal error: ffi.h: No such file or directory 14 | #include <ffi.h> | ^~~~~~~ compilation terminated. Definitely missing the -I for libffi.
Here it is mine: cd _build/default/src/ctypes-foreign && /usr/bin/x86_64-pc-linux-gnu-gcc -O2 -fno-strict-aliasing -fwrapv -pthread -fPIC -march=native -O2 -pipe -fomit-frame-pointer -D_FILE_OFFSET_BITS=64 -fdiagnostics-color=always -I/usr/lib64/libffi/include -g -I /usr/lib64/ocaml -I /usr/lib64/ocaml/bigarray-compat -I /usr/lib64/ocaml/integers -I /usr/lib64/ocaml/stdlib-shims -I /usr/lib64/ocaml/threads -I ../ctypes -o ffi_call_stubs.o -c ffi_call_stubs.c
try src_compile() { strace -f dune build @install --verbose --profile release -j 1 \ 2> ctypes.strace } You should find something like "/usr/bin/pkg-config", "--cflags", "libffi" take note of the pid and wait for changes. Some line above you should see "-I/usr/lib64/libffi/include\n"
Hi, Generated file (as expected) is quite big, not going to upload unless specifically requested. plastiekpoot [07:12:49] /var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/work/ocaml-ctypes-0.23.0 # grep pkg-config ctypes.strace [pid 589] newfstatat(AT_FDCWD, "/var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/work/ocaml-ctypes-0.23.0/_build/install/default/bin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/usr/lib/portage/python3.12/ebuild-helpers/xattr//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/usr/lib/portage/python3.12/ebuild-helpers//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/usr/local/sbin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/usr/local/bin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/usr/sbin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/usr/bin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/sbin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/bin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/opt/bin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/usr/lib/llvm/18/bin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/usr/lib/llvm/17/bin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) [pid 589] newfstatat(AT_FDCWD, "/etc/eselect/wine/bin//usr/bin/pkg-config", 0x7ffd76de45d0, 0) = -1 ENOENT (No such file or directory) No matter how I look at that, it looks broken. So my analysis op how we got to pid 589: [pid 26] clone(child_stack=NULL, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLDstrace: Process 589 attached , child_tidptr=0x7f1c61286e50) = 589 Lots of setup and then: [pid 589] execve("config/discover.exe", ["config/discover.exe"], 0x55ff262b0400 /* 280 vars */ <unfinished ...> pid 26 seems to be the top-most process in this case: clone3({flags=CLONE_VM|CLONE_FS|CLONE_FILES|CLONE_SIGHAND|CLONE_THREAD|CLONE_SYSVSEM|CLONE_SETTLS|CLONE_PARENT_SETTID|CLONE_CHILD_CLEARTID, child_tid=0x7f1c60bad990, parent_tid=0x7f1c60bad990, exit_signal=0, stack=0x7f1c603ad000, stack_size=0x7fff80, tls=0x7f1c60bad6c0} => {parent_tid=[27]}, 88) = 27 rt_sigprocmask(SIG_SETMASK, [INT QUIT TERM CONT TSTP WINCH], NULL, 8) = 0 mmap(NULL, 8392704, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0strace: Process 27 attached ) = 0x7f1c5fbac000 [pid 26] mprotect(0x7f1c5fbad000, 8388608, PROT_READ|PROT_WRITE) = 0 [pid 27] rseq(0x7f1c60badfe0, 0x20, 0, 0x53053053 <unfinished ...> Since pid=27 is the child here, 26 has to be topmost (initially strace didn't output the pid XXX portions of the output). So back to configure.exe: plastiekpoot [07:25:06] /var/tmp/portage/dev-ml/ocaml-ctypes-0.23.0/work/ocaml-ctypes-0.23.0 # ldd ./_build/default/src/ctypes-foreign/config/discover.exe linux-vdso.so.1 (0x00007fa5595f8000) libm.so.6 => /lib64/libm.so.6 (0x00007fa5593d3000) libc.so.6 => /lib64/libc.so.6 (0x00007fa559200000) /lib64/ld-linux-x86-64.so.2 (0x00007fa5595fa000) So nothing strange there, but looking at strings, I cannot find any of the PATHs it looks at in the binary. From _build/log: 109 $ (cd _build/default && /usr/bin/ocamlopt.opt -w -40 -g -I src/ctypes-foreign/config/.discover.eobjs/byte -I src/ctypes-foreign/config/.discover.eobjs/native -I /usr/lib64/ocaml/dune-configurator -I /usr/lib64/ocaml/dune-configurator/csexp -intf-suffix .ml -no-alias-deps -o src/ctypes-foreign/config/.discover.eobjs/native/dune__exe__Discover.cmx -c -impl src/ctypes-foreign/config/discover.ml) 119 $ (cd _build/default && /usr/bin/ocamlopt.opt -w -40 -g -o src/ctypes-foreign/config/discover.exe /usr/lib64/ocaml/unix.cmxa -I /usr/lib64/ocaml /usr/lib64/ocaml/dune-configurator/csexp/csexp.cmxa /usr/lib64/ocaml/dune-configurator/configurator.cmxa src/ctypes-foreign/config/.discover.eobjs/native/dune__exe__Discover.cmx) 126 $ (cd _build/default/src/ctypes-foreign && config/discover.exe) Inspecting default/src/ctypes-foreign/config/discover.ml it seems to look for pkg_config (last looked at Prolog ML more than 20 years back, so rusty at best) and then has some fallbacks. The other two "sources": plastiekpoot [07:33:20] ~ # equery belongs /usr/lib64/ocaml/dune-configurator/csexp/csexp.cmxa /usr/lib64/ocaml/dune-configurator/configurator.cmxa * Searching for /usr/lib64/ocaml/dune-configurator/csexp/csexp.cmxa,/usr/lib64/ocaml/dune-configurator/configurator.cmxa ... dev-ml/dune-configurator-3.16.0-r1 (/usr/lib64/ocaml/dune-configurator/csexp/csexp.cmxa) dev-ml/dune-configurator-3.16.0-r1 (/usr/lib64/ocaml/dune-configurator/configurator.cmxa) Looking at all files in /usr/lib64/ocaml/dune-configurator I still don't get where those odd paths are coming from, what's the possibility that something somewhere is messing with the PATH variable prior to trying to compile ocaml-ctypes?
I see /usr/bin//usr/bin/pkg-config Maybe you have some definition of PKG_CONFIG=/usr/bin/pkg-config or similar around as the executable path is the $PATH concatenated to /usr/bin/pkg-config I don't know if that is accepted, but that is not work with this package actually
(In reply to Tupone Alfredo from comment #13) > I see /usr/bin//usr/bin/pkg-config > > Maybe you have some definition of PKG_CONFIG=/usr/bin/pkg-config or similar > around > > as the executable path is the $PATH concatenated to /usr/bin/pkg-config > > I don't know if that is accepted, but that is not work with this package > actually I was able to replicate with PKG_CONFIG=/usr/bin/pkg-config ebuild ocaml-ctypes-0.23.0.ebuild clean compile
The bug has been referenced in the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=88128d8561ad1f2d6fdf2a96aad24116e3f8a144 commit 88128d8561ad1f2d6fdf2a96aad24116e3f8a144 Author: Alfredo Tupone <tupone@gentoo.org> AuthorDate: 2024-12-21 11:33:14 +0000 Commit: Alfredo Tupone <tupone@gentoo.org> CommitDate: 2024-12-21 11:33:43 +0000 dev-ml/dune-configurator: add 3.17.1 Bug: https://bugs.gentoo.org/946135 Signed-off-by: Alfredo Tupone <tupone@gentoo.org> dev-ml/dune-configurator/Manifest | 1 + .../dune-configurator-3.17.1.ebuild | 39 ++++++++++++++++++++++ 2 files changed, 40 insertions(+)
Can use this version of dune.configurator (3.17.1) to build with that PKG_CONFIG
jkroon@plastiekpoot /etc/portage $ grep -C 1 -r PKG_CONFIG bashrc-# Some bug in (at least) the htop ./configure script. bashrc:export PKG_CONFIG=/usr/bin/pkg-config bashrc- Well, there we go. Good call. That went into our bashrc with the following commit: commit f6b60ceabc8f1803a2dc9f5a06b81e1a1e5f8ab5 Author: Jaco Kroon <jaco@uls.co.za> Date: Tue Jan 10 15:51:58 2017 +0200 Work around braindeadness in htop configure script that doesn't automatically pick up the pkg-config command. There may be others too. And *that* is something with the environment :). With *only* removing that one export line I've managed to remerge ocaml-ctypes, and htop works correctly, so I've reverted that commit from our own tree as well. I *suspect* the new dune-configurator package checks for an absolute path in PKG_CONFIG ... so that'll give it too, thank you for the effort, really thankful you managed to track down the root cause.
