cc @viric @vcunat @shlevy @matthewbauer
was disabled in 6ac7b19
Also
https://www.reddit.com/r/NixOS/comments/8eiw2s/id_like_to_get_ada_compiler_gnat_available_and/

@GrahamcOfBorg build gnat

The configure script fails in this part

# Test for GNAT.
# We require the gnatbind & gnatmake programs, as well as a compiler driver
# that understands Ada.  We use the user's CC setting, already found, and
# possibly add $1 to the command-line parameters.
#
# Sets the shell variable have_gnat to yes or no as appropriate, and
# substitutes GNATBIND and GNATMAKE.
AC_DEFUN([ACX_PROG_GNAT],
[AC_REQUIRE([AC_CHECK_TOOL_PREFIX])
AC_REQUIRE([AC_PROG_CC])
AC_CHECK_TOOL(GNATBIND, gnatbind, no)
AC_CHECK_TOOL(GNATMAKE, gnatmake, no)
AC_CACHE_CHECK([whether compiler driver understands Ada],
		 acx_cv_cc_gcc_supports_ada,
[cat >conftest.adb <<EOF
procedure conftest is begin null; end conftest;
EOF
acx_cv_cc_gcc_supports_ada=no
# There is a bug in old released versions of GCC which causes the
# driver to exit successfully when the appropriate language module
# has not been installed.  This is fixed in 2.95.4, 3.0.2, and 3.1.
# Therefore we must check for the error message as well as an
# unsuccessful exit.
# Other compilers, like HP Tru64 UNIX cc, exit successfully when
# given a .adb file, but produce no object file.  So we must check
# if an object file was really produced to guard against this.
errors=`(${CC} $1[]m4_ifval([$1], [ ])-c conftest.adb) 2>&1 || echo failure`
if test x"$errors" = x && test -f conftest.$ac_objext; then
  acx_cv_cc_gcc_supports_ada=yes
fi
rm -f conftest.*])

if test "x$GNATBIND" != xno && test "x$GNATMAKE" != xno && test x$acx_cv_cc_gcc_supports_ada != xno; then
  have_gnat=yes
else
  have_gnat=no
fi
])

Apparently it desires that $CC can compile ADA and that doesn't hold? (the gnatbind and gnatmake checks do succeed) I don't know... I tried the gnatboot you added and cc didn't really work there, so perhaps that's the issue here?

$ cc -c hello.adb
gcc: error trying to exec 'gnat1': execvp: No such file or directory

cc is not from gnatboot, but rather from stdenv. I don't know about ${CC}. I couldn't figure out how this worked before.

Oh, the command above actually works if I add ${gnatboot.cc}/bin into $PATH, but the output from gnat build failure remains the same as far as I can see.

I just checked, it seems that originally, CC=gcc and both gcc and cc both point into the gnatboot package. We'll need to figure out some way to have gnatboot override stdenv. I'm not sure if the stdenv C compiler is still needed, but some C will need to be compiled.

NIX_CC still pointed to stdenv though.

Yes, I managed to fish the output from the test command:

gcc: error: conftest.adb: Ada compiler not installed on this system

I expect the build probably wouldn't work just with gnatboot itself.

I checked with stdenvNoCC, it doesn't work.

That's not what I meant. I got

checking for clearerr_unlocked... configure: error: C++ compiler missing or inoperational

(the previously failing test for ADA working did succeed) The gnatboot from Gentoo seems not to support C++ (by looking at the files), but recent gcc versions are partially written in C++ and I suspect they can't automatically bootstrap without a C++ compiler.

By placing gnatboot in front of the PATH in the preConfigure phase, we get a lot further.

a-except.adb:860:42: missing argument for parameter "S" in call to "Image_Integer" declared at s-imgint.ads:39
make[3]: *** [../../gcc-9.1.0/gcc/ada/gcc-interface/Make-lang.in:1033: ada/libgnat/a-except.o] Error 1

Looks like an Ada error in the source, possibly because gnatboot is version 4.1 which might be too old to compile version 9.1.

Have you looked what bootstrap sequence for gnat is used elsewhere?

Applying the same approach to gnat6 leads even further – several minutes of compilation:

/nix/store/15vi6br717ndj6vd1cwqhc9pgvk2i5hy-binutils-2.31.1/bin/ld: cannot find crt1.o: No such file or directo
ry
/nix/store/15vi6br717ndj6vd1cwqhc9pgvk2i5hy-binutils-2.31.1/bin/ld: cannot find crti.o: No such file or directo
ry
collect2: error: ld returned 1 exit status
gnatlink: error when calling /build/build/gcc/xg++
make[3]: *** [../gcc-interface/Makefile:2626: common-tools] Error 4
make[3]: Leaving directory '/build/build/gcc/ada/tools'
make[2]: *** [Makefile:197: gnattools-native] Error 2
make[2]: Leaving directory '/build/build/gnattools'
make[1]: *** [Makefile:11308: all-gnattools] Error 2

Commit 8bd9463.

Looks like this still needs to be ported https://github.com/NixOS/nixpkgs/blob/b249f7f1a81765321d65c1f7861598ee4a5020a6/pkgs/development/compilers/gcc/gnat-cflags.patch

I ported it and I'm currently investigating other errors. Will push later today.

Just managed to build gnat6 using gcc6, but I will try to make it work using gcc7 for a bit.

I assumed that if the gentoo gnatboot could be used to build some gnat version that's not too old, that one should hopefully be sufficiently new to build gnat9 and hopefully even some future versions. (I tried 6 first for greater ease of modifications ATM due to gcc6 not being in stdenv bootstrap.)

I've repinned this PR and fixed ghdl-llvm by setting LIBRARY_PATH. If Nix has a better solution than explicitly setting LIBRARY_PATH, let me know, otherwise, this PR is ready to be merged in my opinion.

@vcunat it would be nice if we can get this merged before the feature freeze in two days. Do you need me to do anything else? Please let me know.

I'll test the build locally and merge (today), but I'm afraid we're probably a week or two too late for the 19.09 fork-off, as it will be done from the master branch. Still, I haven't kept a close look at the current plans, e.g. gcc8 default still isn't merged and "was planned" (originally).

Hmm, it still does not build for me: dhx6br4f0m3kjnlvmc59ph7cpg54h3sv-ghdl-llvm-0.36.log – perhaps we'll disable that variant for now?

Looks like this is an instance of the intermittent build failure. It worked for me with 4 and with 8 cores. It might work if you try a few more times with --cores 1. Maybe we should just set enableParallelBuilding = false; on ghdl and gnat?

If one derivation is flaky when building in parallel, setting enableParallelBuilding = false would be the right way.

Can you do that and rebase on top of latest staging?

As langAda doesn't default to true, and it mostly builds too, it'd probably be beneficial to have this in nixpkgs.

Btw, I'm not so sure about the state of that gnatboot distfile on the gentoo servers.

If we need to use binaries, what about something like that as bootstrap?
That seems to be what gentoo does nowadays, too.

{ stdenv
, fetchurl
, autoPatchelfHook
, ncurses5 }:

stdenv.mkDerivation {
  pname = "gnats-bootstrap";
  version = "2014";

  buildInputs = [
    ncurses5
  ];

  nativeBuildInputs = [
    autoPatchelfHook
  ];

  src = fetchurl {
    name = "gnat-gpl-2014-x86_64-linux-bin.tar.gz";
    url = "https://community.download.adacore.com/v1/6eb6eef6bb897e4c743a519bfebe0b1d6fc409c6?filename=gnat-gpl-2014-x86_64-linux-bin.tar.gz";
    sha256 = "0cxpxbx3dnq6fsy4k3qil377znv8k13nn3xgfha9jmpm9p4shqw0";
  };

  buildPhase = ":";

  installPhase = ''
    mkdir -p $out
    cp -R . $out

    rm $out/bin/gps_exe
    rm -R $out/lib/python2.7/ $out/share/gdb-7.7/python-2.7.3 $out/lib/gtk-3.0 $out/lib/gps/gdk-pixbuf-2.0 $out/lib/gps $out/lib/libpyglib-gi-2.0-python.so.0.0.0
  '';
}

There are also binaries for a lot of architectures, more than on the gentoo mirrors.

I've rebased this and was able to build a coreboot toolchain with Ada support, which is required for building coreboot with native graphics initialization for some boards.
Thanks for the work! My rebased version is available on the feature/gnat branch on my fork.

@Lucus16 first of all, thanks for this PR.
I also rebased it and was able to build the coreboot toolchain with this environment. I would love to see this PR merged. Do you have any plans to continue your work ?

It's been a while since I've worked on this. I'll see if I can rebase on staging and build a few times to see if the intermittent build failure is still present.

I rebased and squashed this. I built gnat9 and ghdl-mcode once and gnat6 a few times and did not encounter failures. I've marked ghdl-llvm as broken for now. I think this is ready to be merged.

@sylv-io I'd appreciate it if you could take over and find someone to merge this. I don't have much energy left to spend on this. I'll be around for questions.

I rebased and squashed this. I built gnat9 and ghdl-mcode once and gnat6 a few times and did not encounter failures. I've marked ghdl-llvm as broken for now. I think this is ready to be merged.

@sylv-io I'd appreciate it if you could take over and find someone to merge this. I don't have much energy left to spend on this. I'll be around for questions.

great thx! it's understandable, that your tired of this.
I will do my best to find some folks, but I'm quite new here.

I just did a bunch of small changes:

• update ghdl to 0.37
• use pname
• quote homepage urls
• use https for homepage urls
• marked ghdl-llvm as unbroken (I was able to build it)
• change commit messages to fit contribution guidelines