Piping to xargs does not handle spaces in paths too well, because it splits
up the paths.
For deleting empty dirs, we also need to do several retries, otherwise
`find` will try to go through the directories after they're deleted.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Some programs that use the Python C API have difficulties finding
cross-compiled Python3 without the pkgconfig, so make sure we have
python3.pc and python-3.7.pc in pkgconfig staging dir.
CircleCI requires a package Makefile change to actually
do the CI check, so bump PKG_RELEASE.
Signed-off-by: Daniel F. Dickinson <cshored@thecshore.com>
python3's lib2to3 would fail in silence if python3 and its packages are installed as compiled .pyc files. Root cause is, in Lib/lib2to3/refactor.py, the function get_all_fix_names only searches '.py' fix names.
Signed-off-by: Nj Hsiong <nj.hsiong@gmail.com>
`setuptools` & `pip` whl files were selected via wildcards, because it was
easier in the beginning.
Also, initially there weren't any PYTHON{3}_{SETUTPTOOLS/PIP}_VERSION
variables. But now since these vars exist, it makes sense to use them,
because we can catch easier (at build) time if Python/Python3 bump these
versions.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
This change bumps Python3 version to 3.7.1.
Patch `002-fix-implicit-dh-free-declaration.patch` is now included in
upstream.
This also fixes CVE-2018-1061.
https://www.cvedetails.com/cve/CVE-2018-1061/
Compile & run-tested on x86.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
The linker option -rpath is required to find libs in staging_dir. Now it
is included when building host modules. Without it the import test of
the _ctypes and _uuid modules would fail. The _ctypes module uses
libffi.so.6 from staging, but OpenSUSE LEAP 15 has libffi.so.7.
It will also fail on LEAP 42.x, Fedora28 and 29 and future or old
versions of Ubuntu.
Fix needed in master and 18.06 branches.
Signed-off-by: Jan Kardell <jan.kardell@telliq.com>
No idea how this creeped up. Probably OpenSSL been has updated recently.
Will send this patch upstream as well, but in the meantime we should fix
the Python3 build.
Build error seems to be:
```
<openwrt>/build_dir/target-i386_pentium4_musl/Python-3.7.0/Modules/_ssl.c:4000:5: error: implicit declaration of function 'DH_free'; did you mean 'lh_free'? [-Werror=implicit-function-declaration]
DH_free(dh);
^~~~~~~
lh_free
cc1: some warnings being treated as errors
Python build finished successfully!
The necessary bits to build these optional modules were not found:
_tkinter _uuid nis
To find the necessary bits, look in setup.py in detect_modules() for the module's name.
The following modules found by detect_modules() in setup.py, have been
built by the Makefile instead, as configured by the Setup files:
_abc atexit pwd
time zlib
Failed to build these modules:
_ssl
Makefile:618: recipe for target 'sharedmods' failed
```
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
zlib is now a host tool and the zlib/host package was removed. this
dependency is not needed any more as there will always be a zlib host
library.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Report https://github.com/openwrt/packages/issues/5638
It was mentioned that this causes build failures on Mac OS X.
The default behavior [in the setup.py script] is to check whether
`--with-system-ffi` is present in the CONFIG_ARGS env var.
However that back-fires a bit when `--with-system-ffi=no`, because the
condition `not '--with-system-ffi' in sysconfig.get_config_var("CONFIG_ARGS")`
evaluates to true.
This is a small bug in the `setup.py` script, but it looks like the
easiest/cleanest way to address it on our end is to just remove it entirely
from the HOST_CONFIGURE_ARGS.
At least that's how it looks like when testing on a Linux machine.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
This change was introduced in commit 1c54e2b0fb to address build
issues on Ubuntu 12.04.
However it was reported to cause issues on Mac OS X.
Report: https://github.com/openwrt/packages/issues/5310
It was also reported that removing this on MacOS X fixes the issue.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Since `lang/python` is it's own folder of Python packages
(for both Python 2 & 3), and these build rules are needed
in a lot of packages [especially Python packages],
putting them here makes sense architecturally,
to be shared.
This also helps get rid of the `include_mk` construct
which relies on OpenWrt core to provide, and seems
like a broken design idea that has persisted for a while.
Reason is: it requires that Python 2/3 be built to provide
these mk files for other Python packages,
which seems like a bad idea.
Long-term, there could be an issue where some other feeds
would require these mk files [e.g. telephony] for
some Python packages.
We'll see how we handle this a bit later.
For now we limit this to this feed.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
The .mk snippets are not really usable at the moment, as they cannot be
considered for metadata collection (package DUMP) when included through
include_mk. Python packages do not use include_mk anymore for this reason,
so the install commands can be removed as well.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
The Modules/getbuildinfo.c allows the use of DATE and TIME
macros to be defined via CFLAGS.
These vars, control the build date & time when the
interpreter is opened, and can be read via the
`platform._sys_version()` function.
So, a conversion from SOURCE_DATE_EPOCH to DATE & TIME
is required at build-time.
This is especially needed for `platform._sys_version()`
to work.
The installation of pip seems to rely on this.
The logic has been adapted from:
https://wiki.debian.org/ReproducibleBuilds/TimestampsProposal#Makefile
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
This reverts commits 4333d1dcbf and
074d2863be, making Python packages
discoverable again by pkg_resources.
Fixes#5361.
Signed-off-by: Jeffery To <jeffery.to@gmail.com>
This reverts commit 3c6d14021e.
( which is a revert of commit c764f77dc1 )
The initiall commit ( c764f77dc1 )
was reverted, becase zlib did not have a host-build.
Now it does:
cbe71649bc
So, now it should be good to put this in.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Re-worked patch `003-do-not-run-distutils-tests.patch`
to reduce patch-size.
Removed `011-fix-ncursesw-definition-colisions.patch`
it is fixed upstream.
Refreshed with `make package/python3/refresh`
Resetting PKG_RELEASE to 1.
This variable was never used for pip3 & setuptools, since
VERSION is specified in the package definitions.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
See:
https://github.com/openwrt/packages/issues/5278
This should make Python & Python3 packages reproducible
when building.
In my local tests, I got the same sha256 for a sample
.pyc file, so likely this is the solution that should address
this.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
This reverts commit c764f77dc1.
The commit caused warnings to be displayed at make defconfig etc.
WARNING: Makefile 'package/feeds/packages/python/python/Makefile'
has a host build dependency on 'zlib/host' but
'package/libs/zlib/Makefile' does not implement a 'host' build type
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
This should fix the zlibmodule build on the host side.
Usually, if zlib is not found, Python/Python3 builds fine
without it, but there are some cases where the Python/Python3
interpreter on the host-side requires zlib to run.
At the moment, zlib does not have a host-build.
This should be available when this PR gets merged:
https://github.com/lede-project/source/pull/1329
[ or a similar one that contains host-build support for zlib ].
In the meantime, this change can go into Python/Python3.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
It was reported via
https://github.com/openwrt/packages/pull/5122#issuecomment-347395472
that if bluez-libs is selected as an installable package,
then the error below will show up:
```
* satisfy_dependencies_for: Cannot satisfy the following dependencies for python-light:
* bluez-libs *
* opkg_install_cmd: Cannot install package python-light.
```
This looks like a limitation in the design of package deps,
and maybe a misuse of conditional deps (i.e. PACKAGE_bluez-libs:bluez-libs).
So, to fix this, an idea we're adding an extra symbol
that enfoces installation of bluez-libs if selected.
We also need to add a way to disable bluetooth build
if PYTHON(3)_BLUETOOTH_SUPPORT is de-selected.
Otherwise, bluetooth is installed and the socket
module is broken due to linker errors.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
This should hopefully fix the Python3 build on buildbot.
For a while I assumed it may be a build-bot issue, but
then looking through the packages repo [and finding
the bluez package] it looks like, if you try
to build all packages, Python3 detects the bluetooth
headers installed by bluez.
It looks like Python's bluetooth support was somewhat
broken ; it was not detecting the <bluetooth/bluetooth.h>
header, so a backport from Python3 to Python fixed that.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
That way some python packages can choose
to keep their egg-info dirs, if they want to, or they're needed.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Not sure how it can happen that the files are not
installed via the host build.
Maybe some SDK-like build.
Let's make sure they are installed via InstallDev rule too.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
The host pip install should have the host's CFLAGS, LDFLAGS, etc
available.
And not the target's flags.
Otherwise, weird things can happen when installing
packages (host-side) that need to build C code.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
The host pip install should have the host's CFLAGS, LDFLAGS, etc
available.
And not the target's flags.
Otherwise, weird things can happen when installing
packages (host-side) that need to build C code.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
To install Python packages host side, that
may be needed for a build.
The intent, is to try to reduce host-side Python
packages being installed via LEDE/OpenWrt build system.
Because those seem like a pain to maintain.
The idea is adapted from Yousong's `python-packages`
package.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Depending on execution order the `python-package-install.sh`
script would return a non-zero err code.
So, this enforces that all commands in the script
don't fail (via the `set -e` directive).
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Similar to LEDE/OpenWrt's Build/Compile/Default rule,
and other similarities like this.
This should allow Python packages to define
PyBuild/Compile rules to do specific stuff per
package.
The advantage of using these (over just overriding
Build/Compile) is the VARIANT mechanism that is
in place to support packaging both for Python & Python3.
So, PyBuild/Compile will get picked up for the Python
variant build, and Py3Build/Compile will get picked
up for the Python3 variant build.
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
If you build python/python3 and later decide to build
python(3)-setuptools and/or python(3)-pip, the build won't
re-run without adding `CONFIG_PACKAGE_python(3)-setuptools`
and `CONFIG_PACKAGE_python(3)-pip`.
Seems to resolve issue:
https://github.com/openwrt/packages/issues/4529
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Seems it's called underlinking that's happening only
on Ubuntu 12.04 with libressl (that comes from LEDE's
tools folder).
Link here:
https://ubuntuforums.org/showthread.php?t=1870586
I'm still reading about this a bit.
Since I don't really get it.
But applying that fix (as in the link) seems to fix compiling
on Ubuntu 12.04, and tried also on 16.04 (to make sure).
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
The Build/InstallDev rule is activated only for target builds.
But if someone needs only the host Python, then
these files need to be installed in this phase, and not Build/InstallDev
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Alexandru Ardelean
Daniel F. Dickinson
Nj Hsiong
Jeffery To
Jan Kardell
Karol Wrona
Martin Strobel
Hauke Mehrtens
Matthias Schiffer
Hannu Nyman
Mirko Vogt