Jeffery To [Fri, 6 Mar 2020 22:59:29 +0000 (06:59 +0800)]
seafile-server: Update to 7.1.2, revamp package
* Package scripts that are shipped by upstream in their binary download
* Includes setup scripts (setup-seafile.sh to use SQLite,
setup-seafile-mysql.{sh,py} to use MySQL) and control scripts
(seafile.sh, seahub.sh)
* Does not include seafile-admin, which is not shipped in upstream's
binary download. Combined with the fact that it hasn't been updated
to Python 3 suggests the script has been abandoned.
* Replace previous init scripts with a simplified script
* Previous init scripts (seafile.init, seahub.init) were modified from
older versions of seafile.sh and seahub.sh, but they haven't kept up
with changes to upstream's scripts
* New init script (seafile-server.init) start/stops both Seafile and
Seahub (there is no need to control them separately) by calling
upstream's control scripts
* Replace previous package config file with new config file
* Options in previous config file (seafile.conf) were mainly for using
Seahub in FastCGI mode. FastCGI was deprecated in Django 1.7 and
removed in 1.9; upstream's control script will only start Seahub
using Gunicorn. (Options for Gunicorn including port number can be
changed by editing /etc/seafile/conf/gunicorn.conf.py.)
* New config file (seafile-server.config) has one option that controls
where the Seafile/Seahub data directories are stored
* Patch scripts/binaries to use standard, system-wide directory
locations
* Script files (wrappers for binaries) in /usr/bin
* Binaries (not meant to be run directly by the user) in /usr/libexec
* Config files in /etc/seafile
* Pid/socket files in /var/run/seafile
* Logs in /var/log/seafile
* Include a new script to create the first admin account
* With upstream's original scripts, the user is required to
interactively create the first admin account when Seahub is started
for the first time
* The user will now use the new script (create-seafile-admin.sh) to
create the first admin account after setup (using setup-seafile.sh
or setup-seafile-mysql.sh) and before starting Seafile/Seahub
* seahub.sh is patched to only check if there is at least one admin
account and exit with an error if there is no admin account
* Remove build config options and add seafile-server-fuse package
* The console option controls whether the console window is shown when
Seafile server is run on Windows. It has no use on Linux.
* The fuse option controls whether seaf-fuse is built. (seaf-fuse is a
FUSE implementation that allows the Seafile database/file system to
be mounted to a local directory.) seaf-fuse is now always built and
is available in a separate package (seafile-server-fuse).
Jeffery To [Thu, 13 Feb 2020 14:57:07 +0000 (22:57 +0800)]
seafile-ccnet: Update to 7.1.2, add Python 3 bindings package
* Update package to 7.1.2
* Add Python 3 bindings package. The Python dependency is moved from
seafile-ccnet to python3-seafile-ccnet.
* Refresh patches. The path to valac does not need to be patched in
010-Makefile.patch because $(STAGING_DIR_HOSTPKG)/bin is already in
the path during package build.
* Use SEAFILE_UCI_LOG_DIR and SEAFILE_UCI_SOCKET_DIR, which will be
added to seafile-server
* Install binaries to /usr/libexec instead of /usr/bin, to mirror an
upcoming change in seafile-server
Jeffery To [Wed, 5 Feb 2020 21:34:55 +0000 (05:34 +0800)]
libsearpc: Update to 3.2.0, add Python 3 bindings package
This also moves the Python dependency from libsearpc to python3-searpc,
and adds myself as a maintainer.
Patches:
* 001-fix-memory-leak-GH-48.patch: This is (at this time) the only
difference between the tags v3.2.0 and v3.2-latest.
* 002-no-future-import.patch: The future package is only needed for
Python 2 compatibility; all of the functions/symbols it provides are
available in Python 3.
Jeffery To [Fri, 21 Feb 2020 20:45:28 +0000 (04:45 +0800)]
django1: Fix byte-compiled db migrations not loaded
This patches Django to load byte-compiled (.pyc) db migration scripts,
since Python scripts are often distributed in byte-compiled form in
OpenWrt packages.
Brian J. Murrell [Fri, 13 Mar 2020 13:58:10 +0000 (09:58 -0400)]
mwan3: Ping IPv6 hosts using address not interface
Pinging IPv6 hosts using an interface as a source specifier seems
troublesome. See https://bugs.openwrt.org/index.php?do=details&task_id=2897
for more detail.
Use the desired source interface's IP address instead.
Patches require a bit of rework:
- dropped 010-missing-header.patch
- add 010-do-not-run-test-progs.patch
- re-updated 020-filterh-use-host-built-version.patch
- Makefile.in gets regenerated from PKG_FIXUP:=autoreconf
- added PKG_FIXUP:=autoreconf ; it's easier to patch automake stuff that
big configure scripts
Russell Senior [Tue, 10 Mar 2020 00:20:39 +0000 (17:20 -0700)]
sed: disable acl
If libacl is built, gnu sed finds it during configuration and enables support
linking in libacl. This results in build failures due to the missing dependency.
Consequently, use CONFIGURE_ARGS to disable acl support.
Reported-by: Rosen Penev rosenp@gmail.com Signed-off-by: Russell Senior <redacted>
Karel Kočí [Mon, 9 Mar 2020 13:59:29 +0000 (14:59 +0100)]
btrfs-progs: fix check for btrfs in btrfs scan
The previous implementation always succeeded so no scan was performed.
This now fixes that and it correctly scans for BTRFS devices if BTRFS
support is in kernel.
Yousong Zhou [Mon, 9 Mar 2020 01:41:29 +0000 (09:41 +0800)]
exfat-nofuse: rename to kmod-fs-exfat0
exFAT fs driver has been upstreamed as a staging driver since linux 5.4.
The package name kmod-fs-exfat is used for that.
Three things happened in this commit
- The package will only be built for kernel 4.14 and 4.19
- The driver will be available as opkg package "kmod-fs-exfat0".
"kmod-fs-exfat" is to be used for the one in upstream kernel
- Mark "kmod-fs-exfat0" as deprecated in package title
Signed-off-by: Paul Blazejowski <paulb at blazebox.homeip.net> Signed-off-by: Yousong Zhou <redacted>
(rename, rewording, and deprecation)
- Bump version also in the config file
- Move logread to /sbin location instead of /usr/sbin
Because package logd in OpenWrt's ubox Makefile uses /sbin folder and LuCI has
the same path hardcoded in
modules/luci-base/root/usr/share/rpcd/acl.d/luci-base.json and using
different location results that LuCI is not able to load log data as it
is not found.
* The Makefile creates two packages:
- chicken-scheme-interpreter, same that was created
with the previous version of the Makefile (except
that this one includs Chicken 5.2.0);
- chicken-scheme-full, which includes the compiler,
debugger, profiler, and module manager.
The -full package has some peculiarities:
It is somewhat large, although that is not really
an issue, since it depends on gcc, which already needs
more than 100Mb space.
The binaries are not stripped, because stripped binaries
seem to break the compiler (generated objects are not
properly linked). Since the package is large and depends
on an even larger one, this shouldn't be a problem.
Florian Eckert [Thu, 5 Mar 2020 10:50:00 +0000 (11:50 +0100)]
keepalived: fix building with disable libipvs dynamic
Backport of upstream staging commit to fix compilation issue with x86_64 as
ppc64le build hosts.
https://github.com/pqarmitage/keepalived/commit/42c020aab6e51d09f22a2e4a33ce6fd73009e2dc
Rosen Penev [Sat, 29 Feb 2020 21:57:40 +0000 (13:57 -0800)]
linknx: convert to procd
Removed options UCI parameter. It's not terribly useful. Most of the
parameters can be replaced with procd functionality. procd also demands
processes to run in the foreground.
Removed -fno-builtin. It seems to be a legacy option.
git.99rst.org / openwrt-packages.git / log