| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
Gentoo-Bug: http://bugs.gentoo.org/649822
|
|
|
|
|
|
|
|
|
|
|
| |
Apply upstream patch[1] from SA Bug #7208[2].
[1]:https://svn.apache.org/viewvc/spamassassin/trunk/lib/Mail/SpamAssassin/Plugin/URILocalBL.pm?r1=1684653&r2=1684652&pathrev=1684653&view=patch
[2]:https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7208
Closes: https://bugs.gentoo.org/650638
Closes: https://github.com/gentoo/gentoo/pull/7476
Package-Manager: Portage-2.3.24, Repoman-2.3.6
|
|
|
|
| |
Package-Manager: Portage-2.3.24, Repoman-2.3.6
|
|
|
|
| |
Package-Manager: Portage-2.3.24, Repoman-2.3.6
|
|
|
|
| |
Package-Manager: Portage-2.3.24, Repoman-2.3.6
|
|
|
|
|
| |
Package-Manager: Portage-2.3.24, Repoman-2.3.6
RepoMan-Options: --include-arches="ia64"
|
|
|
|
| |
Package-Manager: Portage-2.3.13, Repoman-2.3.3
|
|
|
|
|
|
|
|
|
|
|
| |
In commit 5248063, I accidentally changed the name of the PostgreSQL
dependency in the OpenRC service script from "postgresql" (correct) to
"postgres" (not so much). This commit changes it back to the correct
name in a new revision. Thanks to Aaron W. Swenson for noticing the
mistake.
Closes: https://bugs.gentoo.org/639916
Package-Manager: Portage-2.3.13, Repoman-2.3.3
|
|
|
|
| |
Signed-off-by: Michael Orlitzky <mjo@gentoo.org>
|
|
|
|
| |
Package-Manager: Portage-2.3.8, Repoman-2.3.3
|
|
|
|
|
| |
Package-Manager: Portage-2.3.13, Repoman-2.3.4
RepoMan-Options: --include-arches="ppc ppc64"
|
|
|
|
|
| |
Package-Manager: Portage-2.3.13, Repoman-2.3.4
RepoMan-Options: --include-arches="hppa"
|
|
|
|
| |
Gentoo-Bug: http://bugs.gentoo.org/636544
|
|
|
|
| |
Package-Manager: Portage-2.3.12, Repoman-2.3.4
|
|
|
|
| |
Package-Manager: Portage-2.3.13, Repoman-2.3.4
|
|
|
|
|
|
|
| |
Rolf Eike Beer)
Package-Manager: Portage-2.3.13, Repoman-2.3.4
RepoMan-Options: --include-arches="sparc"
|
|
|
|
|
| |
Package-Manager: Portage-2.3.13, Repoman-2.3.4
RepoMan-Options: --include-arches="ia64"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In the previous revision (r18), support was added for running spamd as
root through OpenRC. That was done using a new variable called
SPAMD_RUN_AS_ROOT, defaulting to false. The choice to use a boolean
variable -- as opposed to e.g. SPAMD_USER -- was made because passing
"root" as the username to spamd kills it. Thus, SPAMD_USER=root would
not have worked, and we instead test SPAMD_RUN_AS_ROOT to decide
whether or not to specify a username/groupname at all.
The same exact issue arises with systemd; however, systemd offers no
way for us to test the value of SPAMD_RUN_AS_ROOT and act on the
result! The SPAMD_USER proposal is dead in the water for the same
reason, so a different approach was needed. The simplest thing that
could work was to move the "--username" and "--groupname" flags out
of the systemd service file, and into the SPAMD_OPTS variable. That
way, users who know what they are doing can simply drop those flags.
Closes: https://bugs.gentoo.org/635790
Package-Manager: Portage-2.3.8, Repoman-2.3.3
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The r13 revision of spamassassin came with a new init script (and
systemd service file) that runs spamd as the "spamd" user by default,
and that choice is not configurable. There is however a legitimate use
case for running spamd as root; namely, when local users have their
configurations or bayes databases stored in their home directories on
the local filesystem.
This new revision adds back the ability to run spamd as root, through
the SPAMD_RUN_AS_ROOT variable in the OpenRC service configuration
file. This should suffice for the users who have reported problems so
far, and a similar fix for the systemd service is on its way.
The pkg_postinst phase of the ebuild alerts users to the new
configuration variable.
Bug: https://bugs.gentoo.org/635790
Package-Manager: Portage-2.3.8, Repoman-2.3.3
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This new revision adds a patch (thanks to Mark Wright) that fixes the
build with the new openssl-1.1 branch. Our patch is based on the fix
contributed upstream by Noah Meyerhans (thanks Noah) in SA bug 7361.
In an unrelated change, the SRC_URI was updated to use the https
protocol.
Bug: https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7361
Closes: https://bugs.gentoo.org/624858
Package-Manager: Portage-2.3.8, Repoman-2.3.3
|
|
|
|
|
|
|
| |
Rolf Eike Beer)
Package-Manager: Portage-2.3.13, Repoman-2.3.4
RepoMan-Options: --include-arches="sparc"
|
|
|
|
| |
Package-Manager: Portage-2.3.11, Repoman-2.3.3
|
|
|
|
| |
Package-Manager: Portage-2.3.12, Repoman-2.3.3
|
|
|
|
|
| |
Package-Manager: Portage-2.3.12, Repoman-2.3.3
RepoMan-Options: --include-arches="ppc64"
|
|
|
|
|
| |
Package-Manager: Portage-2.3.12, Repoman-2.3.3
RepoMan-Options: --include-arches="ppc"
|
|
|
|
|
| |
Package-Manager: Portage-2.3.12, Repoman-2.3.3
RepoMan-Options: --include-arches="ia64"
|
|
|
|
| |
Signed-off-by: Michael Orlitzky <mjo@gentoo.org>
|
|
|
|
| |
Package-Manager: Portage-2.3.6, Repoman-2.3.1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
As part of bug 590346, I accidentally broke the systemd service file
by starting the service as spamd:spamd when it needs to bind to port
783. Even when the service is run in the foreground, it needs to be
started as root; it will then drop privileges if given the --username
and --groupname flags. The new service file gets this right, with
root owning the main process and spamd:spamd the children.
Thanks to Dan Goodliffe for the report and help debugging the issue.
Gentoo-Bug: 590346
Package-Manager: Portage-2.3.6, Repoman-2.3.1
|
|
|
|
| |
Package-Manager: Portage-2.3.6, Repoman-2.3.1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Add upstream bug URL to patch
( https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7462 )
- Removed -r15 from package.mask
- Unrequested: Renamed patch file now that there is an
upstream bug to match the naming convention of the
other patches
Gentoo-Bug: 627678
Package-Manager: Portage-2.3.6, Repoman-2.3.1
Signed-off-by: Michael Orlitzky <mjo@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add "use lib '.'" calls before "use SATest" calls in tests that are
missing them. This fixes tests that fail when the PERL_USE_UNSAFE_INC=0
env var is set.
Also mark this new package as masked, same as -r14.
Gentoo-Bug: 627678
Package-Manager: Portage-2.3.6, Repoman-2.3.1
Signed-off-by: Michael Orlitzky <mjo@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
| |
This commit drops the StandardOutput=syslog and StandardError=syslog
lines from the new systemd service file. The defaults for those two
options are now "journal" which will eventually hit syslog anyway if
the user so desires. Consequently, we drop After=syslog.target too.
Thanks to grawity on Freenode for the tip.
Package-Manager: Portage-2.3.6, Repoman-2.3.1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit brings the systemd service and its configuration back in
sync with OpenRC:
* The SPAMD_OPTS passed to the daemon now match OpenRC's.
* We no longer pass --daemon to spamd, and thus we don't have to
specify that the daemon forks or track its PID file in the
service.
* The "spamd" user and group are used for the service.
Gentoo-Bug: 590346
Package-Manager: Portage-2.3.6, Repoman-2.3.1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit adds an upstream patch to fix a test suite crash with
newer versions of perl. It also largely rewrites the init script:
* Don't conditionally "use mysql" or "use postgres" in the
OpenRC init script. Even if SpamAssassin is installed without
e.g. MySQL support, if MySQL is present on the system, a user
might store his SpamAssassin data in it.
* Use a declarative style and drop the custom start() and stop()
functions.
* Run the spamd daemon as the "spamd" user.
* Use /run/spamd.pid for the PID file to avoid a "checkpath".
Some updates were made to the ebuild to facilitate those changes:
* Don't modify the init script for USE=mysql or USE=postgres.
* Create a "spamd" user in pkg_preinst().
Package-Manager: Portage-2.3.6, Repoman-2.3.1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In bug 617950, Deniss Gaplevsky asked that we add Redis support for
the SA bayes engine. I looked into it, and it turns out that the
support is already built-in. That makes things a little weird, because
it would have put us in the position of adding a "redis" USE flag that
does nothing.
Some background will hopefully clarify: when I took over as maintainer
of SA, you were required to choose one of USE="berkdb mysql postgres
sqlite". In other words, bayes support wasn't optional, and you had to
choose a database -- there was no "bayes" USE flag. I don't use bayes,
so I added the "bayes" flag to let me proceed without selecting a
database type.
If I had known about Redis at the time, a better solution would have
been to do what I just did: simply drop the REQUIRED_USE constraint
forcing you to choose one of USE="berkdb mysql postgres sqlite". That
is the end result of this commit, although I had to get rid of the
"bayes" flag now to make it all work.
Gentoo-Bug: 617950
Package-Manager: Portage-2.3.6, Repoman-2.3.1
|
|
|
|
|
|
| |
Gentoo-Bug: 625316
Package-Manager: Portage-2.3.6, Repoman-2.3.1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A user asked us to enable Redis support for SpamAssassin in bug
617950, but it should already be there. SpamAssassin has its own
"TinyRedis" client, and doesn't pull in any extra dependencies to
enable it.
This is somewhat confusing when you look at the REQUIRED_USE for
SpamAssassin, which suggests that, to enable USE=bayes, you need to
pick a particular database. I only see three solutions, none of which
are great:
1. Introduce a USE=redis flag that does nothing.
2. Try to document the USE=bayes flag better, to make it clear
that you get Redis support even with USE="-bayes".
3. Allow USE=bayes without picking a database driver.
The first I've ruled out for aesthetic reasons, and the third I think
could do more harm than good (to everyone who wants bayes without
Redis). Therefore I've taken the second option, and updated the
documentation for the local USE=bayes flag. It now mentions that you
don't need USE=bayes to get the Redis bayes backend -- it's always
there.
Gentoo-Bug: 617950
Package-Manager: Portage-2.3.3, Repoman-2.3.1
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The first fix in this revision adds a slot-operator dependency on
dev-lang/perl. This should force SpamAssassin to rebuild in response
to future major upgrades of perl (bug 615856).
The second fix includes a backported patch to support newer versions
of dev-perl/Net-DNS. This should avoid the problems experienced with
the URIDNSBL plugin (bug 590338).
Gentoo-Bug: 590338
Gentoo-Bug: 615856
Package-Manager: Portage-2.3.3, Repoman-2.3.1
|
|
|
|
| |
Bug: 611234
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The PERL_MM_USE_DEFAULT variable is used to disable the interactive
prompts in SpamAssassin's build system. However, when I dropped the
perl-module.eclass a while back, I only set the variable where it
seemed to be necessary. Where it is necessary turns out to depend on
the package manager, because different package managers may treat the
controlling terminal and file descriptors differently, and that's how
MakeMaker decides whether or not to prompt you.
That problem manifests in Paludis, and is hopefully fixed by exporting
PERL_MM_USE_DEFAULT=1 sooner, in src_configure(). This change should
only affect the build, so it is made without a new revision.
Gentoo-Bug: 604518
Package-Manager: portage-2.3.0
|
|
|
|
| |
Package-Manager: portage-2.3.0
|
|
|
|
|
|
| |
Package-Manager: portage-2.3.0
RepoMan-Options: --include-arches="ppc"
Signed-off-by: Agostino Sarubbo <ago@gentoo.org>
|
|
|
|
|
|
| |
Package-Manager: portage-2.3.0
RepoMan-Options: --include-arches="ia64"
Signed-off-by: Agostino Sarubbo <ago@gentoo.org>
|
|
|
|
|
|
| |
Package-Manager: portage-2.3.0
RepoMan-Options: --include-arches="sparc"
Signed-off-by: Agostino Sarubbo <ago@gentoo.org>
|
|
|
|
|
|
|
|
|
|
| |
SpamAssassin, being a perl package, installs a ".packlist" file. By
default that file contains paths from ${D}, our temporary installation
directory. This new revision adds a call to perl_fix_packlist() from
the perl-functions eclass, which replaces those incorrect paths with
the correct ones on the root filesystem.
Package-Manager: portage-2.3.0
|
|
|
|
| |
Package-Manager: portage-2.3.0
|
|
|
|
|
|
|
|
|
|
| |
Call perl_delete_localpod to avoid collisions.
Gentoo-Bug: 603338
Package-Manager: portage-2.3.0
Signed-off-by: Michael Orlitzky <mjo@gentoo.org>
|
|
|
|
| |
Package-Manager: Portage-2.3.2, Repoman-2.3.1
|