There does not seem to be any reason to only move man3 pages.
So, move all man pages to a separate output for documentation.
* gnu/packages/tls.scm
(openssl)[arguments]<#:phases>{move-extra-documentation}: Move
all man pages, not only man3.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
* gnu/packages/tls.scm
(openssl)[arguments]<#:phases>{move-extra-documentation}: Use
'rename-file' instead of 'copy-recursively' and
'delete-file-recursively'.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
This doesn't fix anything broken, just for simplifying
the code a little while we're rebuilding the world anyway.
IMHO this makes the code a little more readable.
* gnu/packages/tls.scm (openssl)[arguments]<#:phases>: Don't refer
to the association list 'outputs', use #$output, #$output:doc
and #$output:static instead.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
This allows using this-package-native-input later.
* gnu/packages/tls.scm
(openssl)[arguments]<#:phases>: Make this a G-expression.
(openssl-1.0)[arguments]<#:phases>: Likewise.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
"sitecustomize.py" is a native input, so look it up
in 'native-inputs', not 'inputs'.
* gnu/packages/python.scm (customize-site): Look up "sizecustomize.py"
in 'native-inputs', not 'inputs'.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
TODO: inform upstream about the cross-compilation error.
* gnu/packages/gnupg.scm
(libgpg-error)[arguments]{fix-gen-lock-obj.sh}: Prevent generated
header files from being sprinkled with ‘\c’.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
* gnu/packages/gpg.scm
(libgpgerror)[arguments]<#:phases>{cross-symlinks}: Only
link to src/syscfg/lock-obj-pub.linux-gnu.h if the package
is being compiled for a Linux target. Do not link either
if the architecture is unknown, as the headers vary with
the architecture.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
The time zone database is architecture-independent,
so trying to cross-compile it is pointless!
* gnu/packages/base.scm
(tzdata)[arguments]<#:target>: Set to #f.
(tzdata)[allowed-references]: Only include the "out" output itself,
to make sure no (architecture-dependent) binaries are installed.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
net-base is pure data, so cross-compiling is pointless.
* gnu/packages/admin.scm
(net-base)[arguments]<#:target>: Set to #f.
(net-base)[arguments]<#:allowed-references>: Disallow all
references.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
It behaves similarily to the other target-...? procedures.
The usage of hurd-triplet? / target-hurd? in libgc appears
incorrect to me, as (%current-system) is normally never false.
* gnu/packages/hurd.scm (hurd-triplet?): Move to ...
* guix/util.scm (target-hurd?): ... here, let its argument
default to (%current-target-system) or (%current-system),
and write a docstring.
* gnu/packages/hurd.scm
(hurd-target?, hurd-system?): Use target-hurd? instead of
hurd-triplet?.
* gnu/packages/bdw-gc.scm (libgc): Likewise.
* gnu/packages/cross-base.scm
(cross-libc)[arguments]<#:configure-flags>: Likewise.
(cross-libc)[arguments]<#:phases>: Likewise.
(cross-libc)[arguments]<#:native-inputs>: Likewise.
* gnu/packages/make-boostrap.scm
(%glibc-stripped)[inputs]: Likewise.
Signed-off-by: Mathieu Othacehe <othacehe@gnu.org>
Make fontconfig use directory contents rather than modification time to
determine cache validity (by pretending that mtime is broken).
* gnu/packages/patches/fontconfig-cache-ignore-mtime.patch: New file.
* gnu/local.mk (dist_patch_DATA): Register it.
* gnu/packages/fontutils.scm (fontconfig)[source]: Use it.
[arguments]: Unset SOURCE_DATE_EPOCH for tests.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
This is a followup to df49fe2a13.
* gnu/ci.scm (%bootstrap-packages): Rename to...
(commencement-packages): ... this, and turn into a procedure. Filter
packages that pass 'supported-package?'.
(cuirass-jobs): Adjust accordingly.
python-cython is required to generate the C source for the libyaml
extension, which is required for the `_yaml' module to work. This in
turn lets the `sanity-check' phase succeed.
* gnu/packages/python-xyz.scm (python-pyyaml)[inputs]: Add
python-cython.
Fixes <https://bugs.gnu.org/49459>.
* gnu/packages/patches/coreutils-gnulib-tests.patch: New file.
* gnu/local.mk (dist_patch_DATA): Add it.
* gnu/packages/base.scm (coreutils)[source]: Use it.
Fixes <https://bugs.gnu.org/49417>.
Reported by Maxime Devos <maximedevos@telenet.be>.
* gnu/packages/patches/glibc-cross-objcopy.patch: New file.
* gnu/local.mk (dist_patch_DATA): Add it.
* gnu/packages/base.scm (glibc)[source]: Use it.
Fixes: <https://bugs.gnu.org/49218>.
* gnu/packages/tex.scm (texlive-bin)[#:configure-flags]: Add
"--disable-luajithbtex" on powerpc64le systems.
(texlive-latex-base)[#:phases][build]: Add "mfluajit" to the disabled-formats
list on powerpc64le systems.
This fixes cross-compilation of GCC, as in:
guix build --target=aarch64-linux-gnu -e '(@ (gnu packages gcc) gcc-10)'
* gnu/packages/gcc.scm (gcc-canadian-cross-objdump-snippet): New variable.
(gcc-5)[source]: Use it instead of inline 'snippet'.
(gcc-8)[source]: Add 'modules' and 'snippet'.
(gcc-9)[source]: Likewise.
(gcc-10)[source]: Likewise.
(gcc-11)[source]: Likewise.
* gnu/packages/make-bootstrap.scm (%bootstrap-tarballs)[arguments]: Make
'use-modules' the first expression. Failing that, 'match' would not be
properly expanded when running on Guile 3.0.
Do not match "gccgo". Match "gcov-dump" and "gcov-tool", as well as
"lto-dump" (which was introduced in GCC 10).
* gnu/packages/gcc.scm (custom-gcc)[arguments]: Update regex used for
matching broken or conflicting executables.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Fixes <https://bugs.gnu.org/49113>.
Reported by Maxime Devos <maximedevos@telenet.be>.
Previously, the cross-compiled libc.so linker script would read:
OUTPUT_FORMAT(elf64-little)
instead of:
OUTPUT_FORMAT(elf64-littleaarch64)
This is because glibc 2.33 runs "objdump -f" instead of
"aarch64-linux-gnu-objdump -f" to determine the output format.
* gnu/packages/patches/glibc-cross-objdump.patch: New file.
* gnu/local.mk (dist_patch_DATA): Add it.
* gnu/packages/base.scm (glibc)[source]: Use it.
Revert openssl support enabled in a1dd57ce83,
as GnuTLS is preferred over OpenSSL in Guix.
* gnu/packages/gnome.scm (glib-networking)[arguments](openssl): Remove flag.
[inputs]: Remove openssl.
Commit f08b070019 intended to skip
'test-fnmatch' on both x86_64-linux and i686-linux but it ended up
skipping it on x86_64-linux only.
* gnu/packages/commencement.scm (findutils-boot0)[arguments]: Use
'member' to test whether (%current-system) is "x86_64-linux" or
"i686-linux".
This is a followup to 4796b5d6d1.
Fixes the "fold-available-packages with/without cache" test in
'tests/packages.scm'.
* gnu/packages/commencement.scm (gcc-toolchain-8): Explicitly refer to
GCC-8.
(gcc-toolchain-10): Turn into an alias for GCC-TOOLCHAIN.
Fixes <https://bugs.gnu.org/48825>.
In short, this change adds the hard-coded "--with-long-double-128" configure
option in one place and removes it from two other places. This changes and
simplifies the use of this option for various architectures that start with
the string "powerpc".
* gnu/packages/gcc.scm (gcc-configure-flags-for-triplet): Add a clause for
targets starting with "powerpc64le-" or "powerpc-" which adds the
"--with-long-double-128" option. This causes any package using this procedure
to be built using this new option on these architectures. In particular, this
affects the gcc package and the gcc-final package, in addition to all the
other versions of GCC defined in (gnu packages gcc).
* gnu/packages/commencement.scm (gcc-boot0)[#:configure-flags]: Remove the
code that adds the "--with-long-double-128" configure option for all
architectures starting with "powerpc", since it is now redundant on the
architectures where it is needed. The gcc-boot0 package uses (and adds to) the
gcc package's configure options. This means that the above change in gcc.scm
is sufficient to ensure that the gcc-boot0 package's configure options will
include "--with-long-double-128" on powerpc64le and powerpc architectures.
Additionally, since the option is apparently not required on the big-endian
powerpc64 architecture, this change also has the nice effect of omitting the
option in that case.
* gnu/packages/cross-base.scm (cross-gcc-arguments)[#:configure-flags]: Remove
the code that adds the "--with-long-double-128" configure option for
powerpc64le, since it is now redundant. The cross-gcc-arguments procedure uses
(and adds to) the configure options of its xgcc argument (a package). This
means that regardless of which gcc from gcc.scm is used as the xgcc, the above
change in gcc.scm is sufficient to ensure that the cross-gcc-arguments
procedure's configure options will include "--with-long-double-128" on the
powerpc64le and powerpc architectures.
