insserv (1.12.0-4) unstable; urgency=low
[ Petter Reinholdtsen ]
* New test test_early_all to control that $all only affect start or
stop sequences. Trying to reproduce issue reported in #485307.
[ Kel Modderman ]
* Add 60_all_keyword_start_only.dpatch to prevent Required-Stop: $all
from modifying start link sort order. (Closes: #499721)
* Make test_early_all() test suite function unconditionally fatal,
this is an importan bug fixed by 60_all_keyword_start_only.dpatch.
insserv (1.12.0-3) unstable; urgency=low
[ Kel Modderman ]
* test_fake_loop() and test_fake_loop_reverse() are now unconditionally fatal
tests because they pass with insserv >= 1.12.0.
* Modify 10_nosuse.dpatch to define ISSUSE as -DNOTSUSE instead of unsetting
it, to avoid fooling the build system.
* Purge the rc symlinks in initddir_purge() function of testsuite-common.
* Add a remscript() function to testsuite-common for deleting initscripts
from test init.d/ directory.
* Add test_all_keyword() function to run-testsuite to observe the
semantics of using the $all keyword. It is currently a non-fatal
test.
* Add test_script_in_runlevel() function to expose segfault reported
in #493202.
* Add 40_badboy_segfault.dpatch to fix segfault when scriptname()
function of insserv does not return script name due to script in
runlevel directory not being a symlink (or corrupt). (Closes: #493202)
* Add 50_symlink_in_initddir.dpatch to defend aginst symlinks
installed to /etc/init.d/ that point to other scripts in
/etc/init.d/, causing insserv to fail due to duplicate provides.
(Closes: #485045)
* Extend test_initd_symlink() test suite case to handle cases which
50_symlink_in_initddir.dpatch must handle.
insserv (1.12.0-2) unstable; urgency=low
[ Kel Modderman ]
* Make test_duplicate_provides() test case more realistic by placing
initscripts in init.d dir before registering with insserv. Make sure first
script is registered.
* Add test_deterministic_order() test function to make sure that when two or
more initscripts provide the same facility, the first script actually can
be registered with insserv reliably.
* Add 30_deterministic_order.dpatch to defend against the inconsistent
directory stream sequence returned by readdir(3) under different
circumstances. (Closes: #494514)
insserv (1.12.0-1) unstable; urgency=low
[ Kel Modderman ]
* New upstream version.
- preserve empty runlevel configurations, override adding or removoval of
existing start/top runlevel configration (Closes: #477415, #492526)
* Discard patches applied to or no longer relevant to new upstream:
- 11_more_warnings.dpatch
- 32_debug_option.dpatch
- 33_nosuse_scripts.dpatch
- 40_segfault_virtprov.dpatch
- 41_cleansystem.dpatch
- 42_loopnochangemsg.dpatch
- 50_sign_warning.dpatch
- 51_overwrite_output.dpatch
- 52_shutdown_links.dpatch
- 60_disable_cfgfilter_stat.dpatch
- 62_warn_on_missing_required_fields.dpatch
- 63_warn_on_missing_default_fields.dpatch
- 64_missing_default_fields_fallback.dpatch
* Discard patches not applied in series:
- 32_debug_option.dpatch
- 34_debian_fixedscripts.dpatch
* Refresh patches that still apply to new upstream:
- 10_nosuse.dpatch
- 31_debian_conf.dpatch
* Add an extra call to list_rclinks() in test_newbug_keepoldorder()
function of run-testsuite, after introducing the bug.
* Make test_adding_start() and test_adding_stop() run-testsuite
functions unconditionally fatal on failure.
* Make test_bogus_facility() run-testsuite function non-fatal.
* Mark tests that are currently failing with a comment in run-testsuite.
* Add test_removing_start() and test_removing_stop() to run-testsuite,
showing that insserv will preserve existing links when script header is
unproperly replaced with empty Default-Start or Default-Stop.
* Make test_adding_start() case a more realistic example.
* Make test_incorrect_stopscripts() a fatal test, insserv should not
add shutdown links when none are present.
* Make test_newbug_keepoldorder() test case non-fatal, the fact that it
passed before was not by design but by luck, as per upstream analysis
which is included with test code.
* Handle CFLAGS in debian/rules more consistently, different options were
passed to make in build and install targets, and dpkg now also tampers
with the variable.
* Remove a bunch of overrides for scripts now with valid LSB information in
testing distribution today, or have been removed from archive.
* Add common test suite function for purging init directory, runlevel links
and dependency makefiles.
* Add test_incorrect_startscripts() to replicate bug report described in
#492526, and make sure insserv do not regress in this behaviour.
* Now we have no overrides to install by default, but we may in future,
modify debian/rules to only install overrides from ./debian/overrides/* if
they are present, but not fail otherwise.
* Add test_initd_symlink() test suite function to test if illegal
symlinks in /etc/init.d/ cause problems.
* Bump Standards Version to 3.8.0, add debian/README.source to describe patch
system to conform. No other changes required.
* Add myself to Uploaders.
* Rename 31_debian_conf.dpatch to 11_debian_conf.dpatch, to keep
Debian port patches in tight sequence.
[ Petter Reinholdtsen ]
* Mention initcycle tool from Adam Jensen in README.Debian.
* Adjust test_initd_symlink test to accept the first insertion
and reject the second, as both scripts have the same provide.
-- Ubuntu Archive Auto-Sync <email address hidden> Tue, 04 Nov 2008 21:26:47 +0000
insserv (1.11.0-9) unstable; urgency=low
[ Petter Reinholdtsen ]
* New test case to demonstrate the bug reported in #477415.
* Include empty directories /etc/insserv.conf.d and
/etc/insserv/overrides in the package, to get dpkg to warn about
content there when purging insserv (Closes: #471237).
* Debconf translations:
- Updated German from Erik Schanze (Closes: #479254).
[ Kel Modderman ]
* Add patch 60_disable_cfgfilter_stat to remove stat check from the
callback function that filters the scanning of /etc/insserv.conf.d/ dir
for confiuration file snippets. Instead do stat check in scan_conf().
* New test case to show that two initscripts cannot have the same Provides
field.
* New test case exposing ability to insert a script into runlevel
configuration that depends on virtual facility that is not provided or
does not exist.
* New test case to check that /etc/insserv.conf.d/ can be used to allow one
or more services to define and provide a virtual facility.
* Ensure custom insserv.conf.d/* files are cleaned after
test_insserv_conf_d() in run-testsuite.
* Add test_onlystart() test case to check we can insert scripts with empty
stop runlevel configuration.
* Add test_broken_header() test case to show it seems possible to insert a
script missing Required-{Start,Stop} lsb info fields.
* Add test_no_default_start() to show that it seems possible to insert script
missing Default-Start field.
* Add test_no_default_stop() to show that it seems possible to insert script
missing Default-Stop field, and that links are created in stop runlevels.
* Add 62_warn_on_missing_required_fields and
63_warn_on_missing_default_fields to enable warnings by default about
missing Provides, Required-Start, Required-Stop, Default-Start or
Default-Stop LSB comment keywords.
* Add a note to new patches to relate them to a case function of the test
suite.
* Add 64_missing_default_fields_fallback to ensure sane fallback of
empty runlevel configuration is used for scripts missing Default-
Start or Default-Stop lsb keywords.
* Add test_noprovides_header() to show that it is possible to insert script
missing Provides keyword, and many warnings are exposed as a result.
* Add test_adding_start() to expose that adding start levels uncleanly
modifies on disk configuration. Give test_adding_start and
test_adding_stop test script unique names.
[ Petter Reinholdtsen ]
* Made test_adding_start and test_adding_stop checks non-fatal for
production builds, as this problem is no regression from earlier versions.
-- Ubuntu Archive Auto-Sync <email address hidden> Mon, 26 May 2008 12:54:21 +0100
