| Age | Commit message (Collapse) | Author |
|---|
|
and update comments
Signed-off-by: David P <megver83@parabola.nu>
|
|
Signed-off-by: David P <megver83@parabola.nu>
|
|
Now called db-import-any, it is used to sync -any packages to non-Arch's architectures
The systemd files are based on db-import-keyring
Signed-off-by: David P <megver83@parabola.nu>
|
|
|
|
[ci-skip]
|
|
|
|
|
|
|
|
|
|
make_repo_torrents do it""
When I reverted this commit, I said:
> Since make_repo_torrents will now only place symlinks in TORRENTPOOL, the
> dead symlink removal will now handle it automatically, and db-cleanup needs
> no special knowledge of TORRENTPOOL.
However, that only means that we can *eventually* revert this. There
are still currently a ton of old plain files in TORRENTPOOL that need
cleaned up first.
Consider reverting this when the following comes back empty:
find "${FTP_DIR}/${TORRENTPOOL}" -type f
This reverts commit 840c13924da569fc38f45181b5ac2a059b6b143e.
|
|
These should be errors, but they shouldn't prevent the creation of torrent
files in other directories.
https://labs.parabola.nu/issues/1253
|
|
|
|
|
|
|
|
make_repo_torrents do it"
Since make_repo_torrents will now only place symlinks in TORRENTPOOL, the
dead symlink removal will now handle it automatically, and db-cleanup needs
no special knowledge of TORRENTPOOL.
This reverts commit 3540ddbf4c2dc25903f3d4f23a86dd795ae620e2.
|
|
Place symlinks in TORRENTPOOL (for compatibility with pacman2pacman
20170705.1 and older). Also pace symlinks to the torrent alongside any
symlinks to the package file.
Use dynamic programming to crawl symlinks.
|
|
all repos"
This reverts commit fa2fbd5d0fd120ec6774f6b28bea4dfaf456ab00.
|
|
|
|
|
|
|
|
make_repo_torrents do it
|
|
- Use libremessages:setup_traps() instead of doing it ourselves
- Use $(dirname ... $0)/../ instead of calling dirname twice
- Don't use `&&` or `||` instead of an `if` block
- Compare $CLEANUP_DRYRUN against the string 'true', rather than
executing it
|
|
|
|
|
|
|
|
Rather than it being a separate executable.
|
|
- Don't bother checking which user is running it, that's hacky
- set -e
* actually bail if we get an error
- Make the torrent directory configurable as TORRENTPOOL
|
|
- Add 'torrent.conf' instead of hard-coding the settings
- The second argument is $FTP_BASE, just load that from 'config', rather
than asking for it.
|
|
|
|
|
|
It is a bit similar in spirit to the old 'any-to-ours' program.
When db-import-pkg imports archlinux32-keyring, it is only released to
core-i686, but we need it to be available on all architectures.
Currently, this is a manual process.
So, add db-import-keyring which imports *only* the -keyring packages, and
releases them to [libre] on each architecture.
|
|
|
|
[ci-skip]
|
|
|
|
Work around for:
https://bugs.archlinux32.org/index.php?do=details&task_id=52
|
|
|
|
|
|
instead of pkgbase
This makes it harder to accidentally do the wrong thing and easier to say
"this is exactly what I mean."
When using "pkgname=", it won't vcs_move or vcs_remove the package. Which
is... fine, since if you gave it a pkgname before it was broken anyway, as
they expect a pkgbase. Perhaps they should track which members of a
split package have been moved/removed?
This reverts commit 2d3abfdc8bb035f4ae0b652c1362c68ca10e76e2.
|
|
Commit 2bbf81 promoted the call to sync_pool to outside of the tag loop,
so it bumped sync_pools's inner msg2 to an msg.
Then c9a7db but it back inside of the tag loop, but didn't demote it back
down to an msg2.
|
|
|
|
|
|
|
|
- Drop pointless -0 flag from xargs (overridden it with -d anyway)
- Fix typo in file path (diff vs dif)
- Because of the ARCHPKGPOOL fallback, call sync_pool for each tag
- Don't call sync_pool if the whitelist doesn't exist (is empty)
- Don't call `expac_file ... ${pkgnames[@]}` if pkgnames is empty, it
it would return info on all packages
|
|
|
|
Pool selection can't fail anymore.
|
|
|
|
|
|
|
|
These values should all be the same.
|
|
I know, I'm a hypocrite for violating
# Please try to refrain from adding new variables to this file.
# Instead, create separate ${toolname}.conf files. Only add a
# variable here if multiple tools start needing the option.
When db-import-pkg moves to use db-{move,update,remove} internally, it will
also need to load an upstream-specific config.local.${UPSTREAM} file (the
systemd service already sets Environment=DBSCRIPTS_CONFIG=… appropriately,
in anticipation of this.
This means that the upstream must be specified *twice* when invoking
db-import-pkg; once in the environment, and once in an arg. Right now it
(essentially) disregards the environment setting, but that won't always be
true. To put a point on that, the test suite currently *doesn't* set the
environment differently for each upstream; once it actually uses the
environment, the test suite will "break".
So, I'd like to unify these, so that the upstream only needs to be
specified once.
If we do this as an argument, it will have to get very kludgey with
overriding DBSCRIPTS_CONFIG internally, so don't do that. That leaves
setting the upstream via DBSCRIPTS_CONFIG= as the only sane option.
At that point, it becomes silly to have a tertiary config file
(db-import-${UPSTREAM}.conf) that is referred to by the secondary config
file (config.local.${UPSTREAM}).
|
