by moving the inclusion of buildlink3.mk files outside of the protected
region. This bug would be seen by users that have set PREFER_PKGSRC
or PREFER_NATIVE to non-default values.
BUILDLINK_PACKAGES should be ordered so that for any package in the
list, that package doesn't depend on any packages to the left of it
in the list. This ordering property is used to check for builtin
packages in the correct order. The problem was that including a
buildlink3.mk file for <pkg> correctly ensured that <pkg> was removed
from BUILDLINK_PACKAGES and appended to the end. However, since the
inclusion of any other buildlink3.mk files within that buildlink3.mk
was in a region that was protected against multiple inclusion, those
dependencies weren't also moved to the end of BUILDLINK_PACKAGES.
one or more .entries files, it can set GCONF2_ENTRIES to the names of those
files and they will get (un)registered at (de)installation time.
Installation is handled after .schemas are installed, and uninstallation
before .schemas are deinstalled.
To achieve this, we need to implement an --unload flag in gconftool-2 to
allow clean removal of .entries from the database. This will be fed back
to authors.
Bump PKGREVISION to 1.
2.4.0
- print usage message when no options passed to gconftool
- API docs updates
- fix UTF-8 handling for command line output
- fix reliability of writing out XML
- ensure recursive unsets get synced
- fix flag value that was "0"
- fix recursive preloading
- fix a memleak
- fix some gconftool messages/segfaults
2.3.3
- new gconf_client_notify() API to fake a notification
- fixes to gconftool --dump feature
- implement schema uninstallation in gconftool so packages
can avoid leaving junk
- fix build issues with linc-free ORBit
- new gconf_client_recursive_unset API
- fix gcc 3.3 strict aliasing warnings
2.3.2
- fix build on Cygwin
- don't complain about missing saved_state in syslog
- improve gconf-sanity-check error text
- added --get-type, --get-list-size and --get-list-element
- exported gconf_entry_copy
- cleaned up headers for C++
- added missing GCONF_CLIENT_GET_CLASS
- hackaround fix for 2.3.1 lockfile breakage
* backends/markup-tree.c (markup_entry_get_value): fix a logic error causing
us to deref a null pointer, reported by Brian Alexander Quistorff.
* configure.in: fix gtk-doc version check.
* configure.in: Added Belarusian to ALL_LINGUAS.
* gconf/gconf-internals.c (gconf_value_list_to_primitive_list_destructive):
fix using the wrong type (value->type not value->list_type) in an error
message, #96210.
* gconf/gconf-internals.c (subst_variables): patch from Benn Vosseteig fixing
a crash with a long path to $HOME.
* doc/gconf/gconf.sgml, doc/gconf-1.0.dtd: Some additions to match the full
range of schema elements curently accepted by gconftool. Supposedly complete
now.
* doc/Makefile.am: change the gconf-1.0.dtd install location to
$(datadir)/sgml/gconf which sounds more standards compliant.
* backends/markup-backend.c (resolve_address): add FIXME about.
Changes in the package:
* Honor PKG_SYSCONFDIR, maintaining the database inside it (PLIST do not need
to care about files in etc/gconf.xml.defaults any more).
* Provide a schemas.mk file that is used to automatically register .schemas
files in the database.
* Add a patch to handle schema deinstallation.
Reviewed by wiz.
day, we can easily create an independant package containing it.
Rationale: The GConf2 library itself does not depend on GTK2, nor any of
the main programs, and this removal will shrink the dependancy tree a lot
on other packages.
Bump PKGREVISION to 5.
Avoid dependancy on gtk-doc (as the package ships with pre-generated
documents) and on gnu gettext (not needed any longer). Some PLIST fixes.
Bump PKGREVISION to 4.
