ppp has multiple variants, so selecting one of them introduces a
recursive dependency for any packge selecting it.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Following a discussion on bugs.python.org:
* https://bugs.python.org/issue29708
* https://bugs.python.org/msg313384
It seems that setting a fixed value to PYTHONHASHSEED guarantees that
the bytecodes are generated consistently/in a reproducible manner.
Hopefully, this is the last bit to make Python3 build reproducible.
Tested this locally on a few files [that were not reproducible without
this change].
The PYTHONHASHSEED is only assigned to the host Python/Python3 during
compilation of byte-codes [from python source].
Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
Use ppp-mod-pppoe instead of shipping its own broken plugin
Always use rp-pppoe.so instead of the (not packaged) pppoe user space
implementation
Signed-off-by: Felix Fietkau <nbd@nbd.name>
* properly install config file
* prevent smsd from starting right after installation to prevent loss of data
(make default device unavailable by turning configuration into comments)
* move init script to procd
* update init script priority to sane value
* log to syslog - at the moment via stdout and procd
this is a workaround to force smsd to stay in the foreground
* the init script can't do a modem reset anymore
use an alarmhandler script instead
* add a local patch to make all processes terminate if the main process dies
* take over the package as new maintainer
Signed-off-by: Harald Geyer <harald@ccbib.org>
It was missing in a previous change bab8741 (tree: use
TARGET_{CFLAGS,LDFLAGS} instead of the package default)
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
At the time of this writing, Open vSwitch official website suggests
http://openvswitch.org whose https couterpart uses a self-signed
certificate, but it redirects to http://www.openvswitch.org, which has a
working https equivalent.
Reported-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
- KCONFIG: bridge compatibility was removed since Open vSwitch 1.10.
See Open vSwitch FAQ.md for details
- The module does not depend on kmod-gre, kmod-vxlan
- Use AutoProbe to remove dependecy on specific priority
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
Open vSwitch needs to run python on build machine to make build-time
required files. python-six library is only required by the openvswitch
python library on target machine, not a build dependency.
We override host PYTHONPATH by overriding it in MAKE_VARS. This way we
can remove 0003-override-pythonpath-via-make-vars.patch
This also fixes shebang wrongly pointing to python on host
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
The python library is a standalone unit. Remove dependency on
PACKAGE_openvswitch to allow users to use it with maybe remote
openvswitch services.
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
This subdir contains multiple pid, unix domain socket files. It's a
custom to put them in it's own subdir
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
Quote from Open vSwitch FAQ.md
Open vSwitch userspace should also work with the Linux kernel module
built into Linux 3.3 and later.
Open vSwitch userspace is not sensitive to the Linux kernel version. It
should build against almost any kernel, certainly against 2.6.32 and
later.
The SUPPORTED_KERNEL dependency for openvswitch kernel module only
makes sense when we are building it from the ovs release tarballs
against mainline kernels. Now that we are using the module from vanilla
kernel itself, the dependency does not exist anymore
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
This reverts 5b5659850d.
In hindsight I have to admit I did not correctly understand the
implications of the uclibc++.mk include.
The include allows a package to follow the user's choice regarding which
C++ library should be the standard. Linking against uClibc++ instead of
libstd++ is not a problem when running musl (which is what I had
incorrectly assumed), as both C++ libs are separate packages. And
uClibc++ is a lot smaller than libstd++, which is probably why it is
even the default C++ lib on OpenWrt currently.
Signed-off-by: Sebastian Kemper <sebastian_ml@gmx.net>
Provide configuration items to select individual backends. This
also allows to fine-tune the package dependencies.
In order to address #5637, we make this dep unconditionally
until upstream allow to opt-in/opt-out libavahi support.
Signed-off-by: Michael Heimpold <mhei@heimpold.de>
- Update haproxy download URL and hash
- Update the haproxy homepage
- Add libatomic to the dependencies as 1.8 needs it
- Make USE_REGPARM an x86-only option as this fixes many warnings and does not do much on non-x86 platforms
- Add USE_GETADDRINFO=1 to use getaddrinfo() to resolve IPv6 host names
- Add USE_TFO=1 to enable TCP fast open
- Unbreak CFLAGS, LD and LDFLAGS by adding the missing backslash after $(ADDON)
- Unbreak IGNOREGIT=1 option (typo)
- Rework LDFLAGS and add libatomic
- Add MEDIUM+ patches (see https://www.haproxy.org/bugs/bugs-1.8.4.html)
Signed-off-by: Christian Lachner <gladiac@gmail.com>
- the Lua-support logic was cleaned up to unbreak Lua-support on non-mips(el) targets. Previously, no target had Lua-support.
- mips and mipsel are both known to currently not build with Lua-support enabled => disable both.
- mips64 and mips64el were tested fine with Lua-support enabled.
Signed-off-by: Christian Lachner <gladiac@gmail.com>
Jianhui Zhao
Hannu Nyman
Felix Fietkau
Alexandru Ardelean
Ted Hess
Jemy Zhang
Michael Heimpold
Federico Di Marco
Dirk Brenken
Harald Geyer
Yousong Zhou
Toni Uhlig
Jiri Slachta
Peter Wagner
Sebastian Kemper
Thomas Heil
Christian Lachner
Nikos Mavrogiannopoulos
Bruno Randolf
W. van den Akker