PKGLOCALEDIR and which install their locale files directly under
${PREFIX}/${PKGLOCALEDIR} and sort the PLIST file entries. From now
on, pkgsrc/mk/plist/plist-locale.awk will automatically handle
transforming the PLIST to refer to the correct locale directory.
This is a bugfix release.
* Entirely new icon set, for palettes, gorm, gmodel, nib and the
application.
* Replaced some of the images for the inspectors.
* Changed formatting in some of the inspectors so that they are
easier to navigate.
* Updated version number to match GNOME version.
Changes 2.12.2:
* init: Fix an out-of-bounds error.
* Build from cvs: Look for gmmproc in generic libdir
directory, fixing the build on some 64-bit systems.
* Glade::get_widget(): Show the name of not-found
widgets on the console, because libglade does not seem to
do this anymore (I think it did once.)
* Example: Delete the dialog when we are finished with it.
* Windows build fixes.
* Windows Build:
- Define GLIBMM_DLL when building with mingw32 or cygwin,
because it is needed by gtkmm.
* OptionGroup: Allow default values, by not initializing them all.
* Slight code size reduction by marking private functions as static.
Makedepf90 is a program for automatic creation of Makefile dependency
lists for Fortran source code. Makedepf90 supports MODULE:s,
INCLUDE:s, cpp #include:s, f90ppr $include:s and coco ??include:s and
set-files.
msgfmt, then it should set the following in the package Makefile:
USE_TOOLS+= msgfmt
To deal with message files that use the "msgid_plural" statement,
which isn't supported in NetBSD<=3.x and also in gettext<=0.10.35, we
determine if the built-in "msgfmt" is sufficiently new enough to
understand "msgid_plural". If it isn't, then we use the msgfmt.sh
script to transform the msgid_plural statements to an equivalent
construct that's understood by older msgfmt tools.
The msgfmt.sh script is a straightforward translation of the original
perl script msgfmt.pl script by Julio M. Merino Vidal into shell and
awk, which are more lightweight dependencies than perl.
We remove the USE_MSGFMT_PLURALS bits in gettext-lib/builtin.mk as they
are made obsolete by the new code in mk/tools/msgfmt.mk.
BUILD_USE_MSGFMT is still supported but will be removed in a separate
commit.
vendor-specific locations for scripts installed by Perl modules. By
default, site scripts go into lib/perl5/site_perl/bin and vendor scripts
go into lib/perl5/vendor_perl/bin. This matches the behavior of the
ExtUtils::MakeMaker that is in pkgsrc perl (which has been similarly
patched). This package now doesn't overwrite the "instmodsh" script
installed by the perl package.
Bump the PKGREVISION to 1.
Sat Apr 8 19:33:35 PDT 2006
0.26 release. Major enhancements and internal rewrites.
Please read these notes carefully, as significant changes are
described. In particular, you _cannot_ upgrade to 0.26
without some attention to the migration, especially if you are
working on a project with other people. See UPGRADE for
details of this procedure.
The changes are large enough that there were 3 pre-releases of
this code; the changes that occurred in each can be seen
below. However, for the convenience of those following
releases, all changes since 0.25 will be summarized in these
release notes. There is no need to read the pre-release notes
individually.
Major changes since 0.25:
- The most user-visible change is that the default name of the
monotone binary has changed to 'mtn'. So, for example, you
would now run 'mtn checkout', 'mtn diff', 'mtn commit',
etc., instead of 'monotone checkout', 'monotone diff',
'monotone commit'.
- Similarly, the name of the workspace bookkeeping directory
has changed from "MT" to "_MTN". As workspaces will
generally be recreated when migrating to this release,
this should not cause any problems.
- Similarly, built-in attrs like 'execute' have had 'mtn:'
prepended to their names. For example, executable files
should now have the attr 'mtn:execute' set to 'true' on
them. The migration code will automatically add this
prefix; no user intervention is needed.
- Similarly, the name of the ignore file has changed from
'.mt-ignore' to '.mtn-ignore'. The migration code will
automatically rename this file; no user intervention is
needed.
- Similarly, the recommended suffix for monotone db files is
now '.mtn'.
These changes are all purely cosmetic, and have no affect on
functionality.
- The most developer-visible change is that the data
structure for representing trees has been completely
replaced, and all related code rewritten. The new data
structure is called a 'roster'. You don't really need to
know this name; unless you are hacking on monotone or using
various debug operations, you will never see a roster.
It's mostly useful to know that when someone says something
about 'roster-enabled monotone' or the like, they're
referring to this body of new code.
This change has a number of consequences:
- The textual format for revisions and manifests changed.
There is no conceptual change, they still contain the same
information and work the same way. The formats were
merely cleaned up to correct various problems experience
showed us, and allow various enhancements now and in the
future. However, this change means that a flag-day
migration is required. See UPGRADE for details.
- Directories are now first-class objects. You can add an
empty directory, must drop a directory if you want it to
go away, etc.
- Attrs are now first-class objects. '.mt-attrs' no longer
exists; attrs are now described directly in the manifest,
and changes to them appear directly in revisions. The
migration code will automatically convert existing
.mt-attrs files to the new first-class attrs. If you have
custom attrs, those may require special handling -- if
this is the case, then the upgrader will tell you.
- The merge code has been rewritten completely. The
interface is currently the same (though this rewrite makes
it easier to improve the interface going forward); if you
have found merging in monotone to be easy in the past,
then you will not notice anything different. If you have
run into problems, then the new merger should make your
life substantially simpler. It has full support for
renames (of both directories and files), intelligent
merging of attrs, improved handling of file content
merges. Is the first known merger implementation based on
a provably correct algorithm (the "multi-*-merge"
algorithm), has exhaustive automated tests, and generally
should give accurate, conservative merges.
- The new code is generally faster, though not yet as
fast as it could be.
Netsync changes:
- The default netsync port has changed 5253 to 4691. 4691 is
our official IANA-assigned port. Please adjust firewalls
appropriately.
- Netsync code has also been largely reworked; new code should
provide better opportunities for
- The protocol is incompatible with earlier versions of
monotone. This should not be a surprise, since the data it
carries is also incompatible (see above)...
New features:
- New option --brief to 'annotate', gives somewhat more
friendly output.
- Several enhancements to log:
- New option --next, to display descendent revisions
(rather than ancestor revisions).
- When 'log -r' is given an ambiguous selector, it now just
logs all matching revisions, instead of requiring the
selector be disambiguated.
- New option --no-files.
- New command 'show_conflicts', performs a dry run merge.
- New command 'ls changed'.
- 'rename' (and its alias 'mv') now accept a broader range of
syntax:
mtn rename foo some_dir
-> renames foo to some_dir/foo
mtn rename foo bar baz some_dir
-> moves foo, bar, and baz to some_dir/foo,
some_dir/bar, and some_dir/baz
- New hook 'validate_commit_message', which may be used to
verify that all commit messages meet arbitrary user-defined
rules.
- New option --log, to log monotone's output to a file.
- New option 'drop --recursive', to remove a directory and its
contents in one swoop.
- The root dir may now be renamed. This is a somewhat exotic
feature, but has some interesting uses related to splitting
up or joining together projects; see new commands
'pivot_root', 'merge_into_dir'.
Minor bug fixes:
- 'serve' with no --bind argument should now work on systems
where the C library has IPv6 support, but the kernel does
not.
- Stricter checking on the internal version of filenames to
ensure that they are valid UTF-8.
- If the database is in the workspace, then it is always
ignored.
- Monotone no longer errors out when using a French (fr)
locale with a non-Unicode codeset.
Other changes:
- Packet commands ('rdata', 'fdata', etc.) have been moved to
'automate'.
- Database storage now uses sqlite's blob support; database
files should be ~1/4 smaller as a result.
- Monotone now uses sqlite 3.3; this means that older versions
of the command line client (e.g., an 'sqlite3' command built
against sqlite version 3.2) cannot be used to poke at a
monotone 0.26 database. Solution is to upgrade your sqlite3
program. Hopefully this is irrelevant to most users...
- Translations updated, and 3 new translations added (de, it,
sv).
Reliability considerations:
- This new codebase has received much less testing under real
world conditions than the codebase used in 0.25, simply
because it is newer. It has been in active use for monotone
development since 8 January 2006, and only a small number of
bugs have been found; all bugs found so far have been very
minor, and none stood any danger of corrupting data.
Furthermore, we are much more confident in the theoretical
underpinnings of the new approach than the old, and the test
suite attempts to exhaustively exercise all new code paths.
However, none of this is or can be a substitute for real
world experience. We advise caution in upgrading to this
version of monotone, and suggest that (especially) those who
upgrade aggressively should pay extra attention to the
monotone mailing list before and after doing so.
Wed Mar 29 05:20:10 PST 2006
0.26pre3 release. This release may be considered a "release
candidate", in that while we need to write some tests and make
sure some bugs are fixed, all features are in and we hope that
no further bug fixes will be needed either. It is still a
pre-release for testing. Do not package it. DO NOT USE THIS
RELEASE UNLESS YOU WANT TO BE A DAREDEVIL.
But, PLEASE PLEASE TEST this release. There are some
non-trivial changes since 0.26pre2, and this is your last
chance!
Major changes since 0.26pre2:
- The name of the monotone binary has changed to 'mtn'.
- Similarly, the name of the bookkeeping directory in
workspaces has changed from 'MT' to '_MTN' (if you have an
existing 0.26-line workspace, just rename the MT directory
to _MTN).
- Similarly, the name of the ignore file has changed from
".mt-ignore" to ".mtn-ignore". 'rosterify' will rename
these automatically (if you have already rosterified, you
get to rename them by hand).
- Similarly, the recommended suffix for monotone db files is
now ".mtn".
- We now perform stricter checking to make sure that filenames
are valid UTF-8. It is in principle possible that this
stricter checking will cause histories that used to work to
break; if you have non-ascii filenames, it is strongly
recommended to test with this release.
- Root dir renaming is now supported. See new commands
'pivot_root', 'merge_into_dir'.
- As a side-effect, it is now possible to run 'rosterify' on
histories in which two independent lines of history were
merged.
- The security fix released in 0.25.2 has been forward-ported
to this release; this prevents some security exposure to
people running monotone as a client on case-insensitive file
systems.
Minor change since 0.26pre2:
- Database now uses sqlite blobs for storage; should be ~1/4
smaller.
- New command: show_conflicts, does a dry-run merge.
- New option 'drop --recursive', to remove a directory and all
its contents in one swoop.
- Changes to 'log':
- New option --no-files
- Including merges is again the default (i.e., it now acts
like 0.25, and not like 0.26pre2).
- When 'log -r' is given an ambiguous selector, it now just
logs all matching revisions, instead of requiring the
selector be disambiguated.
- New option --log, to log monotone output to a file.
- Netsync changes:
- Was sending far too much data in some cases; now does not.
- Several bugs that caused it to lock up fixed
- Tweak to allow 'usher' proxy to transparently redirect
based on client's protocol version, to ease migration
between incompatible protocol versions.
- Packet commands have been moved to 'automate'.
- Fixed bugs in 'db kill_rev_locally', should no longer leave
an inconsistent db behind.
- Translation updates
Other projects receiving notable work:
- Monotone's "dumb server" support (repo distribution over
HTTP/FTP/SFTP etc.) has been ported to 0.26, a first command
line version written, etc.
- The 'usher' netsync proxy used for hosting many databases on
a single machine has received significant cleanups, and the
'webhost' project to provide a simple interface to shared
monotone hosting providers has received even more work.
Sat Feb 11 13:32:51 PST 2006
0.26pre2 release. Inching towards 0.26. If you are using
0.25 or earlier, then make sure to read the very important
notes for 0.26pre1, below. In particular, like 0.26pre1, this
is a pre-release for testing. Do not package it. DO NOT USE
THIS RELEASE UNLESS YOU WANT TO BE A DAREDEVIL.
(Though, in fact, in a month of usage, only one bug has been
found in the new history code, and it was both minor and
harmless. It has additionally been fixed.)
Database changes:
- SQLite 3.3.3 has been imported. 3.3 introduces a new database
format that is not backwards compatible with earlier 3.x releases.
New databases will be created using this new format. Existing
databases remain compatible, and are not converted automatically.
Existing databases can be converted by performing a database
vacuum ('monotone db execute vacuum').
New features:
- New hook validate_commit_message -- use to verify that all
commit messages meet arbitrary user-defined rules.
UI improvements:
- rename (and mv) commands now accept a broader range of
syntax:
monotone rename foo some_dir
-> renames foo to some_dir/foo
monotone rename foo bar baz some_dir
-> moves foo, bar, and baz to some_dir/foo,
some_dir/bar, and some_dir/baz
- Print a warning if it looks like a user has made a quoting
mistake on push/pull/sync/serve (windows cmd.exe has
confusing rules here).
- New command "ls changed".
- New option "--next" to log, which displays descendents of
the start revision.
- Updating to an arbitrary revision now works again (as it did
in 0.25 and earlier). This allows one to, for instance,
switch a working copy to another head, or back up to an
earlier version, while preserving uncommitted changes.
- New option --brief to annotate, gives somewhat more friendly
output.
- Fixed bug that made ticker output from netsync inaccurate.
- In 'log', --no-merges is now the default, use --merges to
override.
- If the database is in the working copy, then it is always
ignored.
Bugs:
- 'serve' with no --bind should now work on systems where the
C library has IPv6 support, but the kernel does not.
- Compile fixes for GCC 4.1 pre-releases.
Other:
- Better detection when users have not run "rosterify", and
more helpful suggestions on what to do in this case.
- Documentation, translation, error message,
etc. improvements.
- Updates to contrib/mtbrowse.sh, simple shell-based monotone
interface.
- Updates to many other contrib/ files, mostly to maintain
compatibility with monotone changes.
Sun Jan 8 01:08:56 PST 2006
0.26pre1 release. Massive rewrites, released for shakedown.
This release is also dedicated to Shweta Narayan.
This release includes massive changes compared to 0.25. The
core versioning code has all been replaced with a completely
different mechanism. Data formats and the netsync protocol
have changed in incompatible ways.
Migration to 0.26pre1 or later is irreversible and requires a
flag day for your project. See UPGRADE for details. Note
that we DO NOT recommend upgrading at this time; see below.
If you have been following the development list for the last
few months, you may have heard about "rosters" -- this is the
name for the new core data structure we use. While the code
is completely different, the user experience should not be
very different. You will never see a roster, unless you are
debugging monotone itself; everything still revolves around
revisions, manifests, and certs.
While this new code has extensive tests, because of these
incompatibilities, it has never been used for real work. The
purpose of this release is to make a version available for the
monotone developers to begin using for day-to-day work, to
shake out bugs.
Let's say that again in caps: THIS CODE IS PROBABLY BUGGY, DO
NOT USE IT IN PRODUCTION UNLESS YOU WANT TO BE A DAREDEVIL.
However, testing of this version with real databases is a good
idea, and we'd very much appreciate hearing about your
experiences.
Some of the many changes:
- New textual format for revisions and manifests; they remain
conceptually the same, but have been tweaked. Manifests
now use the same "basic_io" format as everything else in
monotone, and contain entries for directories, revisions
record file adds slightly differently and record directory
adds for the first time, etc. Because of this format
change, revision hashes are now different; converting
rosters requires a full history rebuild and reissue of certs.
- Directories are now first class. To get rid of a directory
you must remove it; to create a directory, you must add it.
You can add an empty directory.
- Attrs are now first class. The .mt-attrs file is gone;
attributes are now stored directly in the manifest.
- New merge algorithm, based on "multi-*-merge", and more
aggressive, less buggy merge ancestor selection code
- Netsync's core has been largely rewritten. Code is now much
clearer and more reliable, and now includes the ability to
resume interrupted partial transfers. The netsync protocol
version number has been bumped, and netsync now runs on the
IANA-assigned port 4691 by default.
- 100% fewer change_set.cc related bugs. 100% more roster.cc
related bugs. But the idea of touching roster.cc does not
terrify people.
Changes:
- new extensions:
mq (manage a queue of patches, like quilt only better)
email (send changes as series of email patches)
- new command: merge (replaces "update -m")
- improved commands: log (--limit option added), pull/push ("-r" works
on specific revisions), revert (rewritten, much better)
- comprehensive hook support
- output templating added, supporting e.g. GNU changelog style
- Windows, Mac OS X: prebuilt binary packages, better support
- many reliability, performance, and memory usage improvements
Overview of Changes from GLib 2.10.1 to GLib 2.10.2
===================================================
* Bugs fixed:
322476 Missing check for .dylib [Anders Carlsson]
332841 Segmentation Fault when %llu is passed to
vasnprintf and HAVE_SNPRINTF is not defined [Michael McDonald]
333098 Add support for write FDs to GIOChannel [Markus Brinkmann]
334519 Memleak in goption.c::parse_short_option [Chris Wilson]
334471 g_parse_debug_string reads beyond buffer [Morten Welinder]
334611 g_option_context_parse() should not set program
name to '<Unknown>' if it is already set [Chong Kai Xiong]
335207 g_main_context_unref calls g_source_destroy_internal
with incorrect arguments [Chris Wilson]
335216 Slight performance gains (GList, GAsyncQueue) [Chris Wilson]
335209 Use of unitialised memory in g_mem_profile [Chris Wilson]
334943 make check FAIL: threadpool-test
336085 g_option_context_new parameter lacks better
explanation [Claudio Saavedra]
335215 Some breakages with GThreadPool [Chris Wilson]
333879 gthread/gthread-win32.c: IsDebuggerPresent needs
'#define _WIN32_WINDOWS 0x0401' [Kazuki Iwamoto]
334440 dlerror() portability issue causes crash on (old)
a.out NetBSD platform [ITOH Yasufumi]
333916 g_timer_elapsed docs should mention that
microseconds may be NULL [Christian Neumair]
334646 goption + error out params [Christian Persch]
335294 Documentation should not reference G_HAVE_GINT64,
as it's deprecated. [Richard Laager]
* Translation updates (dz,et,he,hi,ka,or,pt_BR,ro,tr)
* Fix non-OpenType fonts losing kerning in 1.12.0 [#336026, Denis Jacquerye]
* Fix blurred underlines on Win32 [#332656, Tor Lillqvist]
* Build fix when having both Win32 and FreeType cairo backends available
[#337502, Alexander Larsson]
* Bugs fixed in this release:
334802,336026,332656,337502
RECOMMENDED is removed. It becomes ABI_DEPENDS.
BUILDLINK_RECOMMENDED.foo becomes BUILDLINK_ABI_DEPENDS.foo.
BUILDLINK_DEPENDS.foo becomes BUILDLINK_API_DEPENDS.foo.
BUILDLINK_DEPENDS does not change.
IGNORE_RECOMMENDED (which defaulted to "no") becomes USE_ABI_DEPENDS
which defaults to "yes".
Added to obsolete.mk checking for IGNORE_RECOMMENDED.
I did not manually go through and fix any aesthetic tab/spacing issues.
I have tested the above patch on DragonFly building and packaging
subversion and pkglint and their many dependencies.
I have also tested USE_ABI_DEPENDS=no on my NetBSD workstation (where I
have used IGNORE_RECOMMENDED for a long time). I have been an active user
of IGNORE_RECOMMENDED since it was available.
As suggested, I removed the documentation sentences suggesting bumping for
"security" issues.
As discussed on tech-pkg.
I will commit to revbump, pkglint, pkg_install, createbuildlink separately.
Note that if you use wip, it will fail! I will commit to pkgsrc-wip
later (within day).
There have been a number of API changes and several methods have been
depreciated in this release. Although current GNUstep programs that use
these methods will still work in this version, there are enough changes
that a new library version was considered useful. See the release notes
for more details
New NSStream classes where added to the library.
User defaults were restructured to create the defaults lazily and
can be set to not write to an external file at all, for developers who
wish to use the library as a stand-alone library or in other situations
where using external resources is not desired.
Version 3.81
* GNU make is ported to OS/2.
* GNU make is ported to MinGW. The MinGW build is only supported by
the build_w32.bat batch file; see the file README.W32 for more
details.
* WARNING: Future backward-incompatibility!
Up to and including this release, the '$?' variable does not contain
any prerequisite that does not exist, even though that prerequisite
might have caused the target to rebuild. Starting with the _next_
release of GNU make, '$?' will contain all prerequisites that caused
the target to be considered out of date. See this Savannah bug:
http://savannah.gnu.org/bugs/index.php?func=detailitem&item_id=16051
* WARNING: Backward-incompatibility!
GNU make now implements a generic "second expansion" feature on the
prerequisites of both explicit and implicit (pattern) rules. In order
to enable this feature, the special target '.SECONDEXPANSION' must be
defined before the first target which takes advantage of it. If this
feature is enabled then after all rules have been parsed the
prerequisites are expanded again, this time with all the automatic
variables in scope. This means that in addition to using standard
SysV $$@ in prerequisites lists, you can also use complex functions
such as $$(notdir $$@) etc. This behavior applies to implicit rules,
as well, where the second expansion occurs when the rule is matched.
However, this means that when '.SECONDEXPANSION' is enabled you must
double-quote any "$" in your filenames; instead of "foo: boo$$bar" you
now must write "foo: foo$$$$bar". Note that the SysV $$@ etc. feature,
which used to be available by default, is now ONLY available when the
.SECONDEXPANSION target is defined. If your makefiles take advantage
of this SysV feature you will need to update them.
* WARNING: Backward-incompatibility!
In order to comply with POSIX, the way in which GNU make processes
backslash-newline sequences in command strings has changed. If your
makefiles use backslash-newline sequences inside of single-quoted
strings in command scripts you will be impacted by this change. See
the GNU make manual subsection "Splitting Command Lines" (node
"Splitting Lines"), in section "Command Syntax", chapter "Writing the
Commands in Rules", for details.
* WARNING: Backward-incompatibility!
Some previous versions of GNU make had a bug where "#" in a function
invocation such as $(shell ...) was treated as a make comment. A
workaround was to escape these with backslashes. This bug has been
fixed: if your makefile uses "\#" in a function invocation the
backslash is now preserved, so you'll need to remove it.
* New command-line option: -L (--check-symlink-times). On systems that
support symbolic links, if this option is given then GNU make will
use the most recent modification time of any symbolic links that are
used to resolve target files. The default behavior remains as it
always has: use the modification time of the actual target file only.
* The "else" conditional line can now be followed by any other valid
conditional on the same line: this does not increase the depth of the
conditional nesting, so only one "endif" is required to close the
conditional.
* All pattern-specific variables that match a given target are now used
(previously only the first match was used).
* Target-specific variables can be marked as exportable using the
"export" keyword.
* In a recursive $(call ...) context, any extra arguments from the outer
call are now masked in the context of the inner call.
* Implemented a solution for the "thundering herd" problem with "-j -l".
This version of GNU make uses an algorithm suggested by Thomas Riedl
<thomas.riedl@siemens.com> to track the number of jobs started in the
last second and artificially adjust GNU make's view of the system's
load average accordingly.
* New special variables available in this release:
- .INCLUDE_DIRS: Expands to a list of directories that make searches
for included makefiles.
- .FEATURES: Contains a list of special features available in this
version of GNU make.
- .DEFAULT_GOAL: Set the name of the default goal make will
use if no goals are provided on the command line.
- MAKE_RESTARTS: If set, then this is the number of times this
instance of make has been restarted (see "How Makefiles Are Remade"
in the manual).
- New automatic variable: $| (added in 3.80, actually): contains all
the order-only prerequisites defined for the target.
* New functions available in this release:
- $(lastword ...) returns the last word in the list. This gives
identical results as $(word $(words ...) ...), but is much faster.
- $(abspath ...) returns the absolute path (all "." and ".."
directories resolved, and any duplicate "/" characters removed) for
each path provided.
- $(realpath ...) returns the canonical pathname for each path
provided. The canonical pathname is the absolute pathname, with
all symbolic links resolved as well.
- $(info ...) prints its arguments to stdout. No makefile name or
line number info, etc. is printed.
- $(flavor ...) returns the flavor of a variable.
- $(or ...) provides a short-circuiting OR conditional: each argument
is expanded. The first true (non-empty) argument is returned; no
further arguments are expanded. Expands to empty if there are no
true arguments.
- $(and ...) provides a short-circuiting AND conditional: each
argument is expanded. The first false (empty) argument is
returned; no further arguments are expanded. Expands to the last
argument if all arguments are true.
* Changes made for POSIX compatibility:
- Only touch targets (under -t) if they have at least one command.
- Setting the SHELL make variable does NOT change the value of the
SHELL environment variable given to programs invoked by make. As
an enhancement to POSIX, if you export the make variable SHELL then
it will be set in the environment, just as before.
* On MS Windows systems, explicitly setting SHELL to a pathname ending
in "cmd" or "cmd.exe" (case-insensitive) will force GNU make to use
the DOS command interpreter in batch mode even if a UNIX-like shell
could be found on the system.
* On VMS there is now support for case-sensitive filesystems such as ODS5.
See the readme.vms file for information.
* Parallel builds (-jN) no longer require a working Bourne shell on
Windows platforms. They work even with the stock Windows shells, such
as cmd.exe and command.com.
* Updated to autoconf 2.59, automake 1.9.5, and gettext 0.14.1. Users
should not be impacted.
* New translations for Swedish, Chinese (simplified), Ukrainian,
Belarusian, Finnish, Kinyarwandan, and Irish. Many updated
translations.
A complete list of bugs fixed in this version is available here:
http://savannah.gnu.org/bugs/index.php?group=make&report_id=111&fix_release_id=103
Changes in version 1.11.4:
=========================
* Enhanced documentation: gtk-doc API docs are
now at 100% coverage.
* allow for NULL in-params to AtkImage coordinate APIs.
* New langs: Georgian (ka) Vladimir Sichinava
Oriya (or) Gora Mohanty
Differences include better tracing functionality, and some internal
changes for some functions returning boolean values.
The target now runs again on my big-endian UltrasparcIIi Solaris 9
machine.
Tailor is a tool to migrate changesets between ArX, Bazaar,
`Bazaar-NG`, CVS, Codeville, Darcs, Git, Mercurial, Monotone,
Subversion, and Tla repositories.
This script makes it easier to keep the upstream changes merged in
a branch of a product, storing needed information such as the
upstream URI and revision in special properties on the branched
directory.
etc. Fix the paths to the tools embedded in the prcs binary. Also
set the default PRCS repository to /var/PRCS instead of $HOME/PRCS,
where HOME is the default home directory of the user building the
package.
Bump the PKGREVISION to 3.
location for both emacs and xemacs.
* List the info files directly in the PLIST.
* Use EMACS_MODULES to deal with a possible xemacs-packages dependency.
libgnomeui 2.14.0
-----------------
* Docs
- Update the API docs a lot (Dinoop Tomas)
* Updated and new translations:
- cs (Petr Tomes), de (Hendrik Brandt), hu (Gabor Kelemen),
it (Francesco Marletta), ka (Vladimer Sichinava)
nn (Kjartan Maraas), pl (GNOME PL team), ro (Misu Moldavan),
sv (Daniel Nylander)
libgnomeui 2.13.90
------------------
* Fix critical warnings in gnome-icon-list (Harry Lu)
* Fix permissions on created directories in the file selector
(Federico)
* Portability improvements in the file chooser (Mark)
* Make the authentication dialog thread safe (jylefort at freebsd org)
* Remove some redundant strings (Clytie Siddall)
* Fix typo in the docs (Behdad)
libgnomeui 2.13.3
-----------------
* Don't scale up small images when creating thumbnails (Lucas Rocha)
* Fix critical warning and crash in thumbnailer (Alex Larsson)
libgnomeui 2.13.2
-----------------
* Sync GOption code with latest libgnome (Christian)
libgnomeui 2.13.1
-----------------
* Fix mime type problem with the thumbnailer (Bastien)
* Fix a crasher (Felix Riemann)
* Fix leaks in the GOption code (Christian Perch)
* Thread safety fixes (jylefort at FreeBSD org)
* Warning fixes (Kjartan)
* Leak fixes in the GOption code (Christian Perch)
libgnomeui 2.13.0
-----------------
* Support the icon theme name spec (Dobey)
* Remove some stale code in the file selector (Kjartan)
* Fix a leak in gnome_entry_save_history() (Paolo Borelli) [
* Fix a GnomeCanvas-CRITICAL warning in gnome-icon-item (Federico)
[#323521]
* Scale images before doing a thumbnail to avoid
eating up all ram on very large images (Victor STINNER) [#307885]
* Fix a crash in gnomefileentry (Petr Mladek) [#315549]
* Fix a mem leak (Kjartan)
* Skip hidden pages in GnomeDruid (Ricardo Fernández Pascual) [#89145]
* Remove useless check and warning about (Jens Elkner) [#118532]
* Don't use gnome-keyring on Win32 (Tor Lillqvist)
* Add support for GOption argument parsing (Pawel Sliwowski, chpe) [#307312]
* Focus the first non-empty field in the password dialog (Nate Nielsen) [#318026]
* Convert text from the file entry to filename encoding before returning it.
Patch from Yydzero <yandong.yao@sun.com>, [#321144]
* Translation updates:
* Mahay Alam Khan (bn)
* Miloslav Trmac (cs)
* Marie Lund (da)
* Adam Weinberger (en_CA)
* Francisco Javier F. Serrador (es)
* Ivar Smolin (et)
* Ignacio Casal Quinteiro (gl)
* Reinout van Schouwen (nl)
* Marcel Telka (sk)
* Christian Rose (sv)
* Theppitak Karoonboonyanan (th)
* Albert Fazlí (tt)
* Funda Wang (zh_CN)
What's new in at-spi-1.7.4:
* Fixed dist to include Accessibility_Selector.idl (missing from dist though
in cvs since 1.7.3).
* Bugfixes to EventDetails event support, and fixed event emission for
implementors of SPI_REMOTE_OBJECT.
What's new in at-spi-1.7.0 through 1.7.3:
* Added idl/Accessibility_Selector.idl to the repository (missing in 1.7.0
through 1.7.2, added in 1.7.3)
* DOCS:
All the IDL is now documented with doxygen-compatible docs.
Just run 'doxygen oxyfile' in the IDL directory. (This will
be automated in a future version).
* NEW API:
idl:
Accessibility::Accessible:getAttributes (name/value pair annotation
for all objects)
Accessibility::Accessible:getApplication (retrieves ref to host app)
Accessibility::Component:getAlpha (get transparency/opacity value of component)
Accessibility::Image:getLocale (get POSIX locale for image and
imagedesc)
Accessibility::Text:getAttributeValue (retrieve a single named
attribute value)
Accessibility::Relation:RELATION_PARENT_WINDOW_OF
Accessibility::Role:ROLE_ENTRY, ROLE_CHART, ROLE_CAPTION,
ROLE_DOCUMENT_FRAME, ROLE_HEADING, ROLE_PAGE, ROLE_SECTION (New roles
for complex docs and forms.)
Accessibility::State:STATE_REQUIRED, STATE_TRUNCATED, STATE_ANIMATED,
STATE_INVALID_ENTRY, STATE_SUPPORTS_AUTOCOMPLETION,
STATE_SELECTABLE_TEXT, STATE_IS_DEFAULT (New states, for complex forms)
Accessibility::EventDetails (new, more detailed info marshalled with events)
cspi:
(AccessibleEvent_getSourceName):
(AccessibleEvent_getSourceRole):
(AccessibleEvent_getSourceApplication):
(AccessibleEvent_getSourceDetails):
New methods for interrogating/demarshalling
event details i.e. source's accessible name, role,
and host app.
(Accessible_getAttributes): New, retrieve annotations/attributes
on objects.
(Accessible_getHostApplication): New, retrieve enclosing Application
instance for an object.
(AccessibleImage_getImageLocale): New, retrieve Locale info for an
image.
(AccessibleComponent_getAlpha): New, retrieve alpha value for
an AccessibleComponent (see discussion in idl section above).
* BUGFIXES:
Some compiler fixes from Kjartan Marass.
Some thread-related fixes from Michael Meeks.
Don't allow non-preemptive listeners to pre-empt events! (Bill Haneman).
Major changes in 2.14.0 were:
* Translation updates
* Fix a11y crash
Major changes in 2.13.92 were:
* Translation fixes for EditableLabel context menu
* Build fixes
Major changes in 2.13.91 were:
* Fix exec line for open-with desktop files
* Translation updates
Major changes in 2.13.90 were:
* Translation updates.
Major changes in 2.13.4 were:
* New background setting mode supported
* Add eel_mime_get_availible_mime_types()
* HIG fixes to open with dialog
* Don't show alerts in taskbar
Major changes in 2.13.3 were:
* Remove titles from alerts
* Use GtkComboBox instead of (deprecated) GtkOptionMenu
Major changes in 2.13.2 were:
* Fix translations in alerts
* Code cleanups
Major changes in 2.13.1 were:
* sort by name in application chooser
2.14.0
======
Fixes
- Don't minimize the window when it's on another desktop and it was the last
active window (Vincent) [#331661]
- Fix C89 compliance (Jens Granseuer) [#332866]
Translations
- Rhys Jones (cy)
- Hendrik Richter (de)
- Rajesh Ranjan (hi)
- Raivis Dejus (lv)
- Mugurel Tudor (ro)
- Laurent Dhima (sq)
- Daniel Nylander (sv)
- Maxim Dziumanenko (uk)
2.13.92
=======
Fixes
- Fix window disappearing in workspace switcher after drag and drop (Vincent)
[#317373]
- Fix corrupted drawing of windows (Vincent) [#155502]
- Fix critical warning with action menu of pinned windows (Vincent)
Translations
- Lasse Bang Mikkelsen (da)
- Kostas Papadimas (el)
- Iñaki Larrañaga (eu)
- Gabor Kelemen (hu)
- Aiet Kolkhi (ka)
- Žygimantas Beručka (lt)
- Duarte Loreto (pt)
- Leonid Kanter (ru)
- Woodman Tuen (zh_HK)
- Woodman Tuen (zh_TW)
2.13.91
=======
Misc
- Automate the creation of some files (Torsten Schoenfeld)
- Don't redefine the _ macro for translation (Vincent) [#328621]
- Make the selector widget translations work again (Vincent) [#323181]
Translations
- Vladimir Petkov (bg)
- Miloslav Trmac (cs)
- Priit Laes (et)
- Luca Ferretti (it)
- Takeshi AIHANA (ja)
- Слободан Д. Средојевић (sr)
2.13.90
=======
Fixes
- Fix crash on workspace activation (Elijah, Bart Vanbrabant)
[#327435]
- Include demands-attention/urgent windows unconditionally in
tasklist; this should help users not miss important windows on
other workspaces (Elijah) [#305979]
- Synchronize the right-click-menu with that of Metacity's (Srirama
Sharma, Leena Gunda, Christian Neumair) [#135710]
- Escape window titles used in pango markup, avoids some crashes too
(Vincent Noel, jylefort FreeBSD org) [#306289]
- Correctly catch and handle demands-attention/urgent state changes
for transients of windows in the tasklist (Elijah) [#317541]
- Make the pager Fitts's law compliant when shadow type is none
(Vincent) [#304248]
- Avoid a compiler warning (Kjartan)
Translations
- Jordi Mallach (ca)
- Adam Weinberger (en_CA)
- Lucas Vieites (es)
- Pauli Virtanen (fi)
- Ignacio Casal Quinteiro (gl)
- Ankit Patel (gu)
- Kjartan Maraas (nb)
- Tino Meinen (nl)
- Kjartan Maraas (no)
- Evandro Fernandes Giovanini (pt_BR)
- Theppitak Karoonboonyanan (th)
- Clytie Siddall (vi)
- Wang Jian (zh_CN)
2.13.5
======
Fixes
- Correctly specify the source indication (Elijah)
- Doc generation fixes (Thomas Vander Stichele)
Translations
- Abel Cheung (zh_HK)
2.13.4
======
Features
- Sort window in a group alphabetically (Xavier Claessens, Vincent)
- Change active window when scrolling on the selector (Nigel Tao)
Fixes
- Remove critical warning (Vincent)
- Use virtual desktop size when computing size of pager (Loic Minier, Vincent)
2.13.3
=======
Features
- Add scrolling support to the tasklist (Xavier Claessens)
Misc
- Remove old hack (Vincent)
Translators
- Ivar Smolin (et)
- Kjartan Maraas (nb)
- Kjartan Maraas (nn)
- Kjartan Maraas (no)
2.13.1
=======
Fixes
- Minor code cleanup (Elijah)
Translators
- Erdal Ronahi (ku)
- Ilyas Bakirov (ky)
libgnome 2.14.0
---------------
What's new since 2.13.90
* Updated and new translations:
- cs (Miroslav Trmac), cy (Rhys Jones), da (Ole Laursen),
de (Frank Arnold), et (Ivar Smolin),
hi (Rajesh Ranjan), hu (Gabor Kelemen), it (Francesco Marletta),
ja (Satoru SATOH), nl (Timo Meinen), nn (Kjartan Maraas),
or (Subhransu Behera), ro (Misu Moldovan), ru (Leonid Kanter),
sv (Daniel Nylander), th (Theppitak Karoonboonyan)
libgnome 2.13.90
----------------
What's new since 2.13.7
* Bug fixes
- Initialize all the members of GnomeModuleInfo
(Christian Perch)
- Use GOption instead of popt as default (Christian Perch)
- Add handling of GNOME_VFS_ERROR_INVALID_URI
(Claudio Saavedra)
libgnome 2.13.7
---------------
What's new since 2.13.6
* Bug fixes
- Fix compilation issues due to the missing inclusion
of popt.h (Emmanuele Bassi)
libgnome 2.13.6
---------------
What's new since 2.13.5
* Bug fixes
- Make popt usage deprecated and move to a full GOption
support (Christian)
libgnome 2.13.5
---------------
What's new since 2.13.4
* Bug fixes
- Fix leaks in GOption code (Christian)
- Don't make toolbars detachable by default (Sebastien)
- Build related cleanups (Mark, Kjartan)
* Docs
- Improve docs for GOption handling in GnomeProgram
* Translations
-
libgnome 2.13.4
---------------
What's new since 2.13.3
* Bug fixes
- Fix the broken fix for GnomeProgram. (Gustavo)
(It broke gnome-python)
libgnome 2.13.3
---------------
What's new since 2.13.2
* Bug fixes
- Give feedback from gnome-url if the host can't be found
(Christian Neumair)
- Fix some warnings (Kjartan)
* Translations
- be, bg, cs, en_CA, es, gl, gu, ja, lt, nb, sk, th, zh_CN.
libgnome 2.13.2
---------------
What's new since 2.12.0
* Features
- New show_input_method_menu gconf key (Matthias Clasen)
- Update of the cursors descriptions for the new GTK (Sebastien Bacher)
- New remember_numlock_state gconf key (Jürg Billeter)
- Add GNOME_URL_ERROR_CANCELLED and return it
in gnome_url_show_with_env (Christian Neumair)
* Bug fixes
- Use G_SEARCHPATH_SEPARATOR_S instead of hardcoded ":"(Tor Lillqvist)
- Add support for GOption argument parsing (Pawel Sliwowski, chpe) [#307312]
- Use gnome-vfs on Win32, too (Tor Lillqvist)
* Translation updates:
* Mahay Alam Khan (bn)
* Miloslav Trmac (cs)
* Martin Willemoes Hansen (da)
* Adam Weinberger (en_CA)
* Francisco Javier F. Serrador (es)
* Ignacio Casal Quinteiro (gl)
* Erdal Ronahi (ku)
* Reinout van Schouwen (nl)
* Marcel Telka (sk)
* Christian Rose (sv)
* Paisa Seeluangsawat (th)
* Albert Fazlí (tt)
* Clytie Siddall (vi)
* Funda Wang (zh_CN)
libsexy is a collection of GTK+ widgets that extend the functionality of
standard widgets shc as GtkEntry and GtkLabel by subclassing them and
working around their limitations.
Changes in version 1.11.3:
=========================
* Fix for localization bug #317458 from Kazuzi IWAMOTO.
* Included docs in tarball as appropriate.
Changes in version 1.11.0:
=========================
Added new APIs - minor enhancements to AtkImage,
AtkDocument, AtkObject, AtkComponent, AtkState.
Significant API additions: Added locale attribute to
AtkImage instances, and Document instances.
Added the ability to assign weakly-typed attributes to
AtkObject as name-value pairs; also added similar
name-value pairs to AtkDocument for attributes which are
intended to be valid over the entire document.
Minor documentation fixes (thanks to Bill Abt).
Added 'tt' (Tatar) locale, Simos Xenitellis
location for both emacs and xemacs.
* List the info files directly in the PLIST.
* Use EMACS_MODULES to deal with a possible xemacs-packages dependency.
* KBabel
o Improve loading of Gettext PO files, especially in case of
recoverable or unrecoverable errors
o Settings of the SVN/CVS dialogs are now project-releated (and
not global anymore).
o Improve sending PO file(s) as email.
* Umbrello
o Use horizontal and vertical lines when drawing association
o Import Rose model files (no diagrams yet,
o Automatically fill useful info into the Perl writer heading
template
o Documentation for association roles not saved
o Default data types not added for old Java generator
o Problem reordering methods in classes/interfaces
o Problem with font size computation/word wrap in note widgets
o Custom operations in sequence diagrams become class operations
o Fork/join symbol appears as a black box
o Multiplicity labels positioned incorrectly when moving entities
o Types of entity's attributes are displayed instead of their
names
o Unable to delete entity from entity list in "UML Model" frame
o Interface names not italicized in diagram view
o Cannot Resize Sequence Diagram Synchronous Messages
o Sequence diagram: constructor message only works at 100
percent zoom
o Documentation for associations is not retained
o Crash when importing Python files
o "void" is imported as a class
o Crash when creating a datatype with the same name as a class
o Crash when a non existing data type is used for an argument of
a new method
o Drag'n'drop doesn't put class under mouse cursor when zoom is
not 1:1
o Crash when refusing to rename a class on importing typedef
o Java import fails at abstract methods or interfaces
to pick up portability mods.
Notable changes in this version:
+ now runs on FreeBSD 6.0 (thanks to F. Senault)
+ now runs on Solaris 9
+ Vital Product Data part of INQUIRY SCSI command handling added (to
appease the Solaris initiator, untested as yet, due to Solaris 10
update 1/06 not running on my Ultrasparc hardware).
- Modified cli-generate to add man page and HTML output options, and
more formatting commands.
- Tweaked the hmac_finish prototype to properly indicate that the
midstate is const and unmodified by the function.
* Avoid hardcoding paths with sed when using the preprocessor is much nicer.
* We don't need to explicitly add ${LIBOSSAUDIO} to LIBS since the
configure script already knows to search for -lossaudio.
* Directly list the info files in the PLIST.
Bump PKGREVISION to 2.
that "gmp" is registered as a direct dependency for any package that
includes ghc/buildlink3.mk to get ghc as a build dependency. This is
needed since software built by ghc requires routines from the "gmp"
shared library. This fixes PR pkg/33100.
Remove the workaround in devel/darcs and x11/wxhaskell.
where the base system doesn't provide stubs for non-threaded programs
(which is just NetBSD<=2 for now).
This is absolutely primitive and just sufficient to make MesaLib usable
with nonthreaded programs.
and replace with appropriate references to PKGINFODIR instead.
* Properly account for split info files during installation.
* Move info file listings directly into the package PLISTs.
This fixes info-file-related PLIST problems.
--infodir, --mandir, etc. As a side-effect, this corrects the location
of the installed info files and man pages to be under "gdbada".
* Pass LDFLAGS through to the build via MAKE_FLAGS so that pkgsrc's rpath
flags are passed to the linker.
* Remove deprecated USE_NEW_TEXINFO.
* List info files in the PLIST instead of the package Makefile.
Bump the PKGREVISION to 2.
info files for both emacs and xemacs. Install the info files correctly
by handling the case where there may be split info files. List the info
files in the PLIST. Use EMACS_MODULES where appropriate to replace the
conditional dependency on xemacs-packages. Bump the PKGREVISION.
INSTALL/DEINSTALL script creation within pkgsrc.
If an INSTALL or DEINSTALL script is found in the package directory,
it is automatically used as a template for the pkginstall-generated
scripts. If instead, they should be used simply as the full scripts,
then the package Makefile should set INSTALL_SRC or DEINSTALL_SRC
explicitly, e.g.:
INSTALL_SRC= ${PKGDIR}/INSTALL
DEINSTALL_SRC= # emtpy
As part of the restructuring of the pkginstall framework internals,
we now *always* generate temporary INSTALL or DEINSTALL scripts. By
comparing these temporary scripts with minimal INSTALL/DEINSTALL
scripts formed from only the base templates, we determine whether or
not the INSTALL/DEINSTALL scripts are actually needed by the package
(see the generate-install-scripts target in bsd.pkginstall.mk).
In addition, more variables in the framework have been made private.
The *_EXTRA_TMPL variables have been renamed to *_TEMPLATE, which are
more sensible names given the very few exported variables in this
framework. The only public variables relating to the templates are:
INSTALL_SRC INSTALL_TEMPLATE
DEINSTALL_SRC DEINSTALL_TEMPLATE
HEADER_TEMPLATE
The packages in pkgsrc have been modified to reflect the changes in
the pkginstall framework.
easier to patch the output of autoconf, also create gdb/configure
with the changes from configure.in directly, removing the build
dependency on autoconf.
Overview of changes between 1.11.99 and 1.12.0
==============================================
* Compilations fixes for MSVC. [#333115, Tor Lillqvist]
* More NULL-check against unusable fonts, helps with Win32 fonts without
a Unicode cmap. [Tor]
* Use g_print instead of printf for debugging output in Win32 backend.
[#332855, Tor]
Overview of changes between 1.11.6 and 1.11.99
==============================================
* Fix problem recently introduced that made Win32 backend render boxes
only. [#332538, Hans Breuer]
* Insert '?' chars instead of invalid UTF-8 sequences in
pango_layout_set_text [#33195]
* Bugs fixed in this release:
332167,332538,331995
Overview of changes between 1.11.5 and 1.11.6
=============================================
* Do not export macros PANGO_GLYPH_EMPTY, PANGO_GLYPH_UNKNOWN_FLAG and
PANGO_GET_UNKNOWN_GLYPH publicly; just in the engine/backend interface.
* Fix a few crashers, especially when a font cannot be read. [#314239]
* Change some g_critical's to g_warning, where it's not application's
fault. [#331723]
* Bugs fixed in this release:
330795,331038,329148,331994,331995,331996,314239,331723
Overview of changes between 1.11.4 and 1.11.5
=============================================
* Cleanup work on the win32 backend. Families are not listed all
in lower case anymore.
* New public function: pango_win32_font_description_from_logfont.
* pango-view accepts --wrap now, and lists backends in --help output.
* A couple minor bug fixes.
Overview of changes between 1.11.3 and 1.11.4
=============================================
* Major change in the low-level interface of Pango. The special value
PANGO_GLYPH_EMPTY is now defined to mean a "draw nothing" glyph. This
was what glyph 0 was used for previously. As a result, bumped up the
Pango module version, meaning that modules compiled against earlier
versions are probably not compatible with this release. [#73147]
* New framework in examples for building a binary called pango-view,
that can be used to render a text file using any of the available
backends. This is installed into bindir now.
* Enhancements to the FT2 backend, to draw boxes on missing glyphs.
* Better hexbox drawing for cairo backend: Hinting hexbox, and
drawing a single-row hexbox for very small font sizes.
* Improved documentation.
Overview of changes between 1.11.2 and 1.11.3
=============================================
* Do not crash if the shapers fail.
* Use g_slice for various struct allocations.
* 100% symbol coverage in docs.
* Misc bug fixes and documentation improvements.
* Janitory works, like removing unneeded files.
Overview of changes between 1.11.1 and 1.11.2
=============================================
* Draw hex box on missing glyphs for cairo-fc backend.
* New --with-dynamic-modules option added to configure.
* Use g_slice for various small memory allocations [Matthias Clasen]
* Misc bug fixes, optimizations, and documentation improvements.
Overview of changes between 1.11.0 and 1.11.1
=============================================
* ATSUI support for the cairo backend. [Anders Carlsson]
* Fixed a couple of regressions with font size and scaling.
* Draw a dashed empty box on missing glyphs for cairo-fc backend.
* Support OpenType features for basic (Latin/Greek/...) module.
* Support swash feature in Arabic module.
* Misc bug fixes.
* Misc documentation improvements.
Optimizations:
* Cache GObject private data in PangoFcFont.priv [Federico Mena Quintero]
* Add a constant-sized light-weight cache for glyph extents.
* Use g_slice_* instead of GMemChunk in mini-fribidi.
Overview of changes between 1.10.1 and 1.11.0
=============================================
* Line-breaking algorithm updated to Unicode 4.1.
* Code borrowed from GNU FriBidi resynched to version 1.0.7.
* New improved Tibetan shaper module. [Pema Geyleg]
* Bug fix in Khmer shaper module. [Jens Herden]
* Respect fontconfig reassignment of font pixelsize. [Funda Wang]
* Make OpenType GPOS handling more robust/correct. [Greg Aumann]
* Various documentation fixes and improvements.
* Various gcc warning fixes.
* Misc bug fixes.
Optimizations:
* Add a lazy fixed-size per-font cache to map characters to glyphs,
in the cairo backend. [Federico Mena Quintero]
* Optimize character to script mapping by adding a linear table for
characters < U+2000, and also caching the last mid point in
bsearch. [Federico, Matthias Clasen]
* Enabled mini-fribidi's trashstack instead of direct mallocs.
* Short circuit the bidi algorithm for provably unidirectional text.
* Make mini-fribidi process UTF-8 directly.
* Several strlen and g_utf8_strlen calls removed. [Owen Taylor, Billy Biggs]
* Use call tables in the OpenType layout code.
* Use quarks for accessing GObject data. [Federico]
* Open Pango modules with lazy-bind flags. [John Rice]
* Use new negative offsets in g_utf8_pointer_to_offset to avoid
scanning from the beginning of line on cursor movement.
* Avoid redundant PANGO_IS_FC_FONT checks.
* Windows Build:
- ustring.h: Tag npos with GLIBMM_API, in order to
dllexport it on win32. Bug 332438.
- Updated MSVC++ build files and README, for MS Visual Studio 2005.
* gmmproc code generator:
- WRAP_METHOD() Take an extra optional argument: deprecated depractiontext -
so that we can insert the appropriate doxygen tag in the documentation, where
the C documentation does not do it for us.
package name matches the distname and avoids confusion because there is
no such thing as glade1 any more in pkgsrc (and hasn't been for a very
long time already).
Glade 2.12.1 (Oct 9 2005)
============
o Fixed missing include file for GnomeDB widgets.
o Translation updates.
Glade 2.12.0 (Sep 14 2005)
============
o Added support for new properties - "GtkWindow::urgency_hint",
"GtkAboutDialog::wrap_license", "GtkIconView::reorderable",
"GtkFileChooserDialog::do_overwrite_confirmation",
"GtkFileChooserButton::do_overwrite_confirmation",
"GtkMenuBar::pack_direction", "GtkMenuBar::child_pack_direction".
o Added 12 new named icons in GTK+ and GNOME.
o Fixed bug using "Named Icon" property for widgets other than GtkWindow.
o Fixed code generation for "GtkAboutDialog::destroy_with_parent".
Martijn van Buul.
GNU libsigsegv is a library for handling page faults in user mode. A page
fault occurs when a program tries to access a region of memory that is
currently unavailable. Catching and handling a page fault is a useful
technique for implementing:
* Pageable virtual memory
* Memory-mapped access to persistent databases
* Generational garbage collectors
* Stack overflow handlers
* Distributed shared memory
without underscores (REPLACE.*.old, REPLACE.*.new, and REPLACE_FILES.*).
Also convert REPLACE.*.new= ${SH:Q} back to ${SH}, as it should not be quoted
here, if at all.
Ok with rillig.
Fix structure padding issues on some architectures
Fix compile warnings with newer versions of gcc
Fix memory leak when continuously rescanning for devices
Check for device nodes in /dev/usb under Linux
Fix compile problem with Darwin/Mac OS X port
Add support for Darwin/Mac OS X 10.4
Fix error when transactions are aborted
Requery endpoints when switching alternate settings
Support more BSD based distributions
Fix problem with short control messages in BSD port
Various other minor fixes
Changes since 0.28.2:
added check for whether SSL_read/write takes void * or char * parameter
added _XOPEN_SOURCE to filedescriptorincludes.h for sys/socket for irix
added collapse argument to charstring::split
added configure test for -Wall
switched lots of:
char a[charstring::length(b)];
to:
char *a=new char[charstring::length(b)];
...
delete[] a;
to improve compiler compatibility
removed #ifdef __GNUC__ around static variable definitions, I'll add
#ifndefs for compilers that don't allow them (like SCO's) if
necessary
added -D__EXTENSIONS__ to CPPFLAGS
added charstring::escape/unescape
defaulted zeropadding to 1 rather than 0, I'm not sure if it's a bug
in glibc or not, but printf("%0*lld\n",0,(int64_t)0); prints
nothing rather than 0, but printf("%0*ld\n",0,(int32_t)0);
prints 0. Using 1 instead of 0 for the number of 0's works
in all cases.
solaris's strchr/strrchr return const char *,
so I made charstring::findFirst/findLast return const char *
use snprintf's rather than sprintf's now
uses fchmod/fchown rather than chmod/chown now
made envelope classes out of all classes
added a configure test for ftok that takes a char * argument
switched a bunch of AC_TRY_LINK's to AC_TRY_COMPILE's in configure
script
configure script does less work looking for getpwnam_r and cousins now
if pthread libs can't be found, configure also tries just plain -pthread
added #ifdefs for individual POSIX_FADV_* and MADV_* macros
in configure script, for cygwin, tests for w32api files and functions
are omitted
configure tests for strtoll/strtoull now
configure tests for caddr_t types for all mmap-related functions
individually now
unsupported memorymap methods return false now rather than not existing
added -pthread when compiling as well as linking
fixed a time-remaining-related bug in snooze class
fixed a bug where the date/time string buffer could easily be overrun
Changelog doesn't have entry for 2.06.
2006-01-02 Andreas J. Koenig <andreas.koenig.gmwojprw@franz.ak.mind.de>
* release 2.05
* fix the testcase for recursion so that it compiles and works
also after patch 26370 to perl after which stashes are not
autovivified anymore.
2005-12-25 Andreas J. Koenig <andreas.koenig.gmwojprw@franz.ak.mind.de>
* release 2.04
* Export.pm now strict clean
* Makefile.PL up to date
* added ChangeLog.svn
* added tests for recursion, pod, podcover
* Fixed rt.cpan.org #8766--recursion
Changes since 5.6.4 includes:
* Fixed a small timing hole where the read from the read end of the output
pipe of the server shell can fail. (The read is retried.)
* Added a macro for testing mmap failure to accommodate a picky compiler.
* The minor version number is being bumped. There is a new (albeit tiny)
feature being added to the syntax. See the doc on the pseudo-macro.
* Fixed a bug wherein out-of-order definitions could get mistyped causing
various catastrophic results.
* Definition and license text files were mmap-ed in without checking for
the size being a multiple of a page. strlen() would seg fault.
Horrible, horrible design of mmap!
* disable autoopts/test/getopt.test unless autoopts has been fully installed.
(It requires getopt.tpl which requires a fully installed environment.)
* several other usability improvements to the tear off library.
* template shell code may now use a shell function "die" that will print
an error message and shoot down the master autogen process.
* the tear-off/add-in libopts library will now specify that the
"options.h" header is to be installed.
* realpath(3C) usage in libopts has been cleaned up. It is now only used
if PATH_MAX is defined for the platform. canonicalize_file_name(3C) is
used where available.
0.25.2 release. Important security fix for Windows and OS X
users.
With versions of monotone prior to this release, a person with
commit access could commit a malicious file with a name like
"mt/monotonerc". When anybody else then checked out this
revision on a system with a case-folding filesystem --
usually, this means, "on Windows or OS X" -- then their
monotone would run arbitrary Lua code stored in this file.
The _only_ change in this release as compared to 0.25 is that
the existing checks against files in MT are now extended to
check for mt, Mt, and mT.
All users on Windows and OS X, or otherwise checking out
versioned source on a case-insensitive filesystem, are
recommended to upgrade immediately. Binaries used only for
serving, or only on case-insensitive filesystems (i.e., most
Unix users), are not affected.
(0.25.1 was never released in source form. The original
0.25 build for Windows was found to have problems on NT 4, and
0.25.1 was Windows-only rebuild with NT 4 compatible
libraries.)
Since 6.6:
- Add "int" to Solaris "end" and "etext" declaration in gc.h. Declared
the symbols with underscores and as arrays, since that's what's actually
used. Perhaps this could all just be removed? (Thanks to John Bowman.)
- Fixed ARM GC_test_and_set code. (Thanks to Kazu Hirata and Paul Brook.)
- Added casts for assignments to hb_last_reclaimed, which truncate the
value. Added a cast to GC_adj_words_allocd. Use GetModuleHandleA
when retrieving a handle to kernel32.dll under win32. (Thanks to the
Visual Prolog developers.)
- Added Tandem S-Series support. (Thanks to Craig McDaniel. A modified
version of his patch was applied, and hence breakage is probably not
his fault.)
- Remove spurious gc:: qualifier for operator delete[] in gc_cpp.h.
(Thanks to Hanno Boeck.)
- Changed a test for LINUX in config_macros.h to one for __linux__.
- Fix ppc 64 test_and_set code by removing it. (Thanks to Christian
Thalinger.)
- Add prototypes for GC_finalizer_notifier and GC_thr_init. (Thanks to
David Ayers.)
- Use ld instead of nonexistent ldz instruction in Darwin FindTopOfStack.
(Thanks to Andreas Tobler.)
- Add support for Darwin/X86. (Thanks to Geoff Norton and the Mono
developers.)
- Merge in some recent gcc fixes. Add ppc64 asm code. (Thanks to Bryce
McKinley and other gcj developers.)
- Scan MEM_PRIVATE sections under Windows ME and predecessors.
- Interior pointers with some largish offsets into large objects could
be ignored, if GC_all_interior_pointers was set. (Oddly this worked
correctly for stack references if it was not set. Otherwise it failed
for both stack and heap references.) Thanks to Andrew McKinlay for the
critical test case.
- Integrated Tatsuya Bizenn's NETBSD threads support, with some
minimally tested changes.
- Added GC_strdup and friends to make leak detection work correctly
for strdup clients. (Thanks to Jon Moore.) Fixed the existing strdup
with malloc redirection to handle a null malloc return correctly.
- Fix Makefile.am, so it handles exe extensions under Cygwin correctly
for gctest.
we use a shared sqlite, but new-sqlite can read old-sqlite files for
users of monotone 0.25 which still uses the pre-3.3 internal sqlite.
From the NEWS file:
0.14:
- support new format of monotone db (using BLOBs instead of base64 encoding)
/!\ these newer monotones (after 0.26pre2) use sqlite 3.3, if you're building
monotone-viz with a shared sqlite lib, make sure it is compatible !
- display a nice dialog when the database is locked (e.g. during netsync)
- when a revision has a tag cert, use it as label instead of the revision's id
0.13:
- support monotone 0.26pre1 (it still works fine with monotone <= 0.25)
- stop displaying `disapprove' nodes in a special way (this allows a faster
loading of the database)
* Yet another round of XPDF-related security fixes.
* Mis-detection of man pages as part of TAR archives fixed.
* More Mime-types for the OLE2 extractor. Also ignore (harmless)
libc errors in plugins when extracting.
* More TAR improvements: keywords 'date' and 'format' are
extracted. More checksums variants were added. Long filenames
as produced by GNU and Schilling tar (possibly Solaris pax also)
are extracted.
Changes 0.5.9:
* Made TAR extractor parsing more robust.
* Fixing crash in MIME-extractor due to typo in the code.
* Fixed security problems in PDF extractor
makeinfo if no native makeinfo executable exists. Honor TEXINFO_REQD
when determining whether the native makeinfo can be used.
* Remove USE_MAKEINFO and replace it with USE_TOOLS+=makeinfo.
* Get rid of all the "split" argument deduction for makeinfo since
the PLIST module already handles varying numbers of split info files
correctly.
NOTE: Platforms that have "makeinfo" in the base system should check
that the makeinfo entries of pkgsrc/mk/tools.${OPSYS}.mk are
correct.
developer is officially maintaining the package.
The rationale for changing this from "tech-pkg" to "pkgsrc-users" is
that it implies that any user can try to maintain the package (by
submitting patches to the mailing list). Since the folks most likely
to care about the package are the folks that want to use it or are
already using it, this would leverage the energy of users who aren't
developers.
libgsf 1.14.0
Daniel Nylander:
* Initial Swedish PO translation. [Debian #351382]
Ivan Wong :
* Only export symbols which are in the public headers (Win32).
Jody:
* Fix parsing xml files with &
* bump to 1.14.0 to indicate a stable release.
* improve header versioning.
* Add ODF thumbnail support to gsf-office-thumbnailer
* Enable ODF meta data import/export.
Morten:
* Fix GsfOutputIConv. [#323503]
* Plug leak in GsfInfileZip.
changes:
-debugger improvements
-improved the documentation generator
-support more sourcefile types in editor, as qscintilla does
-fixes/cleanup
-improvements to project management UI
-better shell completion
-added a templates system
-support for Mercurial VCS
changes:
Switched to using mmap as the internal allocation function
Improved the logging of error information about pointers
Added return.h support for ia64 and ARM processors
Added compat and arg_checking of atoi, atol, and memmove
Added backwards compatibility for %d usage in logfile name
Removed force-linear token
-bugfixes
changes:
-Removed the -html20 and -html32 options, the -html output is now
version 4.01
-Removed the -latex209 and -latex2e options, the -latex output is
version 2e
-HTML option generates a cxref stylesheet and uses it
-fixes/cleanup
-Parsing improvements: support some more language features
changes:
faster status, diff, and commit
reduced memory usage for push and pull
improved extension API
new bisect, gpg, hgk, and win32text extensions
short URLs, binary file handling, and optional gitweb skin for hgweb
numerous new command options including log --keyword and pull --rev
improved hooks and file filtering
of postgresql80-client. According to Ralf Wildenhues from libtool,
this fragment is just broken and the pre-1.5.21 was effectively a NOP.
Bump libtool-base revision.
Overview of changes between 1.10.3 and 1.10.4
=============================================
* Fix crasher assertion `split_index > 0' in ellipsization.
1.16 Mon Feb. 6, 2006
- converted to use Module::Build (Rob Kinyon)
- refactored &addChild and &addChildren to be
implemented in terms of &insertChild and
&insertChildren (Rob Kinyon)
- other misc. refactorings (Rob Kinyon)
- updated Scalar::Util version dependency (Stevan Little)
- updated copyrights for the new year (Stevan Little)
charting control flow within the program.
GNU cflow is able to produce both direct and inverted flowgraphs for C
sources. Optionally a cross-reference listing can be generated. Two
output formats are implemented: POSIX and GNU (extended).
Input files can optionally be preprocessed before analyzing.
Overview of Changes from GLib 2.9.6 to GLib 2.10.0
==================================================
* Bugs fixed:
328997 64bit pointer trunction in glib slab-allocator
[Pascal Hofstee]
331110 g_cond_broadcast(inform_cond) without holding
inform_mutex [Chris Wilson, Sebastian Wilhelmi]
332093 Fix some leaks in the tests [Kjartan Maraas]
332435 g_utf8_strlen returns wrong value if a maximum
number of bytes to check is specified
[Matthias Clasen]
331367 gslice requires more POSIX-like semantics for
GPrivate destructors [Tor Lillqvist]
* Documentation improvements [Matthias, Kang Jeong-Hee,
Tor Lillqvist, Stefan Kost]
* Translation updates (el,eu,ka,uk)
Overview of Changes from GLib 2.9.5 to GLib 2.9.6
=================================================
* Bugs fixed:
329124 distclean removes README [Kjartan Maraas, Tim Janik]
317679 GRelation field type not documented [Behdad Esfahbod]
329123 Typo in GTime docs [Kjartan Maraas]
* Documentation improvements [Sven Herzberg, David
Schleef, Kjartan Maraas, Behdad Esfahbod]
* Translation updates (cs,cy,it,ko,pt,sq,sr,sr@Latn,ru
Overview of Changes from GLib 2.9.4 to GLib 2.9.5
=================================================
* Memory management:
Runtime debugging support: The slice allocator
can be turned off by setting G_SLICE=always-malloc
in the environment. Zeroing of freed memory can
now be turned on at runtime by setting
G_DEBUG=gc-friendly in the environment. [Tim Janik]
* Bugs fixed:
328253 HP-UX/IA-64 uses ".so" as default shared library
extension [Albert Chin]
143380 unicode-encoding test fails converting to UTF-16
with libiconv [Marc Moorcroft]
328254 Build breakage (GSlice) [Jens Ganseuer]
328705 C99ism in glib/gmem.c [Kazuki Iwamoto]
* Translation updates (da,et,zh_CN)
Overview of Changes from GLib 2.9.3 to GLib 2.9.4
=================================================
* Type system:
Fix a problem with g_object_compat_control() which
can lead to segfaults in GTK+ applications on 64bit
platforms.
* Thread suppport:
Unused threads now fall back to the global pool after
500 milliseconds, where they wait for another
max-idle-time milliseconds. [Sebastian Wilhelmi]
* Fix a memory allocation problem in GKeyFile. [Morten
Welinder]
Overview of Changes from GLib 2.9.2 to GLib 2.9.3
=================================================
* GTree:
- Replace the simple recursive implementation by
a nonrecursive, threaded one [Maurizio Monge]
* Change g_filename_display_name and
g_filename_display_basename to use the Unicode
replacement character U+FFFD instead of a question
mark, and don't append "(invalid encoding)" [Matthias]
* Documentation improvements [Sven Herzberg, Federico
Mena Quintero, Stefan Kost]
* Bugs fixed:
323937 gslice.c in glib 2.9.1 doesn't build on Mac OS X
[Bogdan Nicula]
326558 Some test failures on IRIX 6.5 [Daichi Kawahata]
169285 "threaded" tree implementation for GTree
[Maurizio Monge]
326747 g_filename_display_basename adds (invalid encoding)
[Alberto Ruiz]
Other contributors: Christian Kellner, Murray Cumming
New and updated translations (bg,ca,de,es,et,gu,ja,nl,th,vi)
Overview of Changes from GLib 2.9.1 to GLib 2.9.2
=================================================
* Memory management:
- Add tests for cache colorization [Tim Janik]
- Minimize space consumption if small amounts of differently
sized slices are allocated, at a small performance cost. [Tim]
* Thread support:
- Add g_atomic_pointer_set() and g_atomic_int_set() [Tim Janik,
Sebastian Wilhelmi]
- Add g_thread_pool_set_sort_function() to allow sorting the
tasks of a threadpool. [Martyn Russell]
- Add g_thread_pool_set_idle_time() to allow unused threads
to exit after a certain time. [Martyn]
* Type system:
- introduce a new type GInitiallyUnowned, which has an initial
floating reference. [Tim]
- Add support for GType parameters. [Matthias]
* Main loop:
- Add g_main_context_is_owner() to determine if the current
thread is the owner of the context. [Michael Meeks]
* Provide g_access(), g_chdir(), g_unlink(), g_rmdir() as
wrapper functions instead of macros. [Manish Singh]
* Documentation improvements [Tim, Matthias, Federico Mena Quintero,
Stefan Kasal, Dan Williams]
* New and updated translations (en_CA,fi,fr,gl,ml,nb,no,zh_HK,zh_TW)
* Bugs fixed:
324179 g_allocator_new() returns pointer to const dummy which Gtk+ 2.8
tries to modify [J. Ali Harlow]
324332 g_option_context_parse() returns false without setting error
[Tim-Philipp Müller]
324950 GLIB 2.9.1 testcase errors [Dan Yefimov]
325015 gslice.c: process.h is needed on Windows [Kazuki Iwamoto]
321978 G_DATALIST_GET_FLAGS() macro is not casting datalist to
gpointer [Andrew Paprocki]
316221 G_LOCK warns about breaking strict-aliasing [Michal Benes,
Stanislav Brabec]
325273 Error in documentation for glib_check_version () [Declan Naughton]
325310 g_spawn_sync hangs when catching both stdout and
stderr [Tor Lillqvist]
325249 gcc warning when using g_rmdir from <glib/gstdio.h> [Jani Monoses]
325864 glib/gthreadpool.c:"#define debug(...)" is C99 [Kazuki Iwamoto]
325874 Should say somewhere that source IDs are > 0 [Dan Williams]
325438 a typo (compatability) [Stefan Kasal]
323937 gslice.c in glib 2.9.1 doesn't build on Mac OS X [Bogdan Nicula]
Overview of Changes from GLib 2.9.0 to GLib 2.9.1
=================================================
* Memory management
- The slice allocator is implemented [Tim Janik]
- g_slice_free_chain() has been renamed to
g_slice_free_chain_with_offset() [Tim, Behdad Esfahbod]
- Mem chunks are deprecated [Matthias Clasen]
* Data structures
- Hash tables are refcounted, and have a boxed type [Tim]
* Thread support
- Support for Solaris threads has been removed
[Sebastian Wilhelmi, Andrew Paprocki]
- g_async_queue_sort(), g_async_queue_push_sorted() have
been added to allow GAsyncQueue to be used as a priority
queue, together with the corresponding _unlocked
variants [Martyn Russell]
* GObject:
- The concept of a floating initial reference has been
moved from GtkObject to GObject [Tim]
* Win32 changes:
- Make g_rename() replace existing files [Tor Lillqvist]
* Misc new API:
- G_GUINT64_CONSTANT macro to define guint64
constants [Andrew Paprocki]
- G_GNUC_WARN_UNUSED_RESULT macro to instruct the
compiler to emit a warning if the value returned
by a function is ignored. [Arjan van de Ven, Alex Larsson]
- GList and GSList now have sort functions which take an
extra user data argument [Martyn Russell]
- g_param_spec_ref_sink() has been added for consistency [Tim]
* $LOGNAME is respected when determining user data. [Laszlo Peter]
* Other changes and bug fixes [Tim, Matthias, Behdad,
Christian Persch, Benedikt Meurer, Andrew Paprocki,
Kazuki Iwamoto, Alexis S. L. Carvalho, Stanislav Brabec,
Andreas Schwab, Kalle Vahlman]
* Documentation
- Deprecation warnings carry version information [Matthias]
- The slice allocator has been documented [Matthias, Tim]
- Other improvements [Morten Welinder]
Overview of Changes from GLib 2.8.x to GLib 2.9.0
=================================================
* Unicode support:
- The Unicode tables have been updated to Unicode 4.1,
adding several new values to the GUnicodeBreakType
enumeration. This breaks Pango <= 1.10
[Behdad Esfahbod]
- The various Unicode character predicate functions
(g_unichar_isalpha, g_unichar_isdigit,...) have
been optimized
[Behdad]
- g_utf8_pointer_to_offset, g_utf8_offset_to_pointer:
These functions handle negative offsets now, and
going backwards in g_utf8_offset_to_pointer uses
"stutter stepping".
[Larry Ewing, Matthias Clasen]
* Memory management:
- Mem chunks are no longer used internally in GLib and
GObject. GMemChunk will be deprecated in GLib 2.10
- All APIs based on GAllocator (g_list_push/pop_allocator,
and similar push/pop_allocator functions for other
data structures) have been deprecated, since they
never worked as intended.
- The g_slice_* functions have been added as a
new API for fast allocation of small memory blocks.
The implementation in GLib 2.9.0 is just a simple
wrapper around malloc. GLib 2.10 will have an
efficient and scalable implementation.
[Tim Janik, Matthias]
* Pattern matching:
- g_pattern_match has been optimized to avoid
unnecessary recursion.
[Tim, Matthias]
* g_intern_string, g_intern_static_string:
- New functions to intern strings. These are now used
by GObject to avoid duplicating static strings
[Matthias]
* g_thread_foreach:
- New function to iterate over all GThreads
[Tim, Matthias]
* g_date_set_time_t, g_date_set_time_val:
- New functions to set a GDate from a time_t or
GTimeVal value. g_date_set_time has been deprecated
in favor of these.
[Roger Leigh]
* g_snprintf and g_vsnprintf:
- These functions are no longer declared in gprintf.h,
since they are in glib.h
[Matthias]
* Replaced many instances of g_return_val_if_fail and g_return_if_fail with
non-asserting logic.
* Fixes many critical warning problems. (Bug #331926 for instance)
file specifications (usually called "file names", but not to be
confused with the contents of a file, or Perl's file handles), such as
concatenating several directory and file names into a single path, or
determining whether a path is rooted.
changes since 1.1.11:
2005-04-04
* patch from Don MacAskill <don@smugmug.com> 'flush_all' doesn't
seem to work properly. Basically, if you try to add a key which
is present, but expired, the store fails but the old key is no
longer expired.
2005-01-14
* Date: Thu, 18 Nov 2004 15:25:59 -0600
From: David Phillips <electrum@gmail.com>
Here is a patch to configure.ac and Makefile.am to put the man page in
the correct location. Trying to install the man page from a
subdirectory results in the subdirectory being used in the install
path (it tries to install to doc/memcached.1). This is the correct
thing to do:
- create a Makefile.am in the doc directory that installs the man page
with man_MANS
- modify Makefile.am in the base directory to reference the doc
directory using SUBDIRS
- modify the AC_CONFIG_FILES macro in configure.ac to output the
Makefile in doc
2005-01-14
* pidfile saving support from Lisa Seelye <lisa@gentoo.org>, sent
Jan 13, 2005
2005-01-14
* don't delete libevent events that haven't been added (the deltimer)
patch from Ted Schundler <tschundler@gmail.com>
2004-12-10
* document -M and -r in manpage (Doug Porter <dsp@dsp.name>)
2004-07-22
* fix buffer overflow in items.c with 250 byte keys along with
other info on the same line going into a 256 byte char[].
thanks to Andrei Nigmatulin <anight@monamour.ru>
2004-06-15
* immediate deletes weren't being unlinked a few seconds,
preventing "add" commands to the same key in that time period.
thanks to Michael Alan Dorman <mdorman@debian.org> for the
bug report and demo script.
Make pkglint happer
This also fixes a number of security issues:
http://www.securityfocus.com/archive/1/425584/30/0/threaded
> Version 2.20.1
> --------------
>
> + Many PostgreSQL fixes, including fixing whine.pl on Pg 8
> (bug 301062) and fixing the --regenerate option of collectstats.pl
> for all versions of Pg (bug 316971). However, users who want full
> PostgreSQL support are encouraged to use the 2.22 series, as
> certain PostgreSQL bugs were discovered that will not be fixed
> in 2.20 (their fixes were too complex).
>
> + In Bugzilla 2.20, the "administrator" user created by checksetup.pl
> would not ever be sent email, because their email preferences were
> left blank. This has been fixed for 2.20.1. However, if you created
> this administrative user with Bugzilla 2.20, make sure to go back
> and enable their Email Preferences. (bug 317489)
>
> + The bzdbcopy.pl script mentioned in these release notes
> has now actually been checked-in to the 2.20 branch, and so
> it's included in this release. (bug 291776)
>
> + When there's only one Classification, you now won't be required
> to pick a Classification on bug entry. (bug 311489)
>
> + You can no longer add dependencies on bugs you can't see.
> (bug 141593)
>
> + The CC list is included in "New" bug emails, again. (bug 313661)
>
> + In the original 2.20, certain scripts were not correctly using
> the "shadow database," if it was specified. This has been fixed
> in 2.20.1. (bug 313695)
>
> + "Saved Searches" that were saved before Bugzilla 2.20, would throw
> an error if they contained "Days Since Bug Changed." as part of their
> criteria. This has been fixed in Bugzilla 2.20.1. (bug 302599)
>
> + You can now successfully delete a product even when Target Milestones
> are turned off. (bug 317025)
>
> + checksetup.pl now correctly pre-compiles templates for languages other
> than English. (bug 304417)
>
> + The "All Closed" chart that is created by default in New Charts
> now actually represents all closed bugs, and not all bugs in the
> product. (bug 300473)
>
> + CSV bug lists with more than 1000 dates now work properly. (bug 257813)
>
> + Various bugs with upgrading from previous versions of Bugzilla
> have been fixed. (bug 307662, bug 311047, bug 310108)
>
> + Many, many other bug fixes. See http://www.bugzilla.org/status/changes.html
> for details on what was fixed between 2.20 and 2.20.1.
appropriately otherwise.
- The server keypair is now stored and read from ${PKG_SYSCONFDIR}/keys
instead of being inside the database.
- Provide and use two files (read-permissions and write-permissions) to
set up netsync's access control.
- During monotone-server-init, run monotone under the unprivileged user
so that it creates files in the correct places (if any).
- Add a note to monotone-server-init to let the user ensure that the
UID and GID are correct. (I always get them wrong otherwise.)
- Make the rc.d script print a "divisory" line in the log file so that
different sections are easy to distinguish.
Bump PKGREVISION to 2.
NEXT.pm adds a pseudoclass named NEXT to any program that uses it. If a method
m calls $self-NEXT::m()>, the call to m is redispatched as if the calling
method had not originally been found.
- Added cli-generate, a program to produce C header and source from
command-line interface description files.
- Modified the crypto HMAC code to do precalculation of the intermediate
hash results.
- Fixed base64 decoding on systems where char is unsigned by default.
Thanks Gerrit Pape.
- Tweaked the ghash_add function to return a pointer to the newly
created entry (since some uses will want to use it immediately).
- Stop blocking SIGSEGV with the sig_all_* functions, as this has
resulted in bugs causing endless signal loops.
PyChecker is a tool for finding bugs in python source code. It finds
problems that are typically caught by a compiler for less dynamic
languages, like C and C++. It is similar to lint. Because of the
dynamic nature of python, some warnings may be incorrect; however,
spurious warnings should be fairly infrequent.
Differences from previous version:
+ when looking for the target name supplied by the initiator during full-phase
login, break out of the loop immediately a match is found.
+ only syslog information if it's available
+ Present multiple targets properly
+ Check that the initiator address is allowed to discover targets.
+ Log a LOG_INFO error if an initiator attempts to discover targets it
shouldn't be.
It requires that your main application is built on top of libevent. Libdnsres'
API essentially mirrors the traditional gethostbyname and getaddrinfo
interfaces. All return values have been replaced by callbacks instead.
The code borrows heavily from the BSD resolver library. In fact, it is an
extremely ugly hack to make the BSD resolver library non-blocking and
thread-safe without changing the API too much.
patch-ab is only an unintrusive short-term fix, discussion with the authors
how to fix it correctly has started.
NetBSD >= 3.0 supports grantpt() but it invalidates the slave
FD (see grantpt(3) on NetBSD) obtained through openpty() so we discard
the (now invalid) descriptor for the slave tty. This causes Tty.xs to open
the slave tty again.
The issue should be really fixed by using posix_openpt() instead of
openpty(). The functions posix_openpt(), grantpt(), unlockpt()
and ptsname() belong together and should be used ahead of
all the other ways to create the master and slave tty, not just
on NetBSD. See also
http://www.opengroup.org/onlinepubs/009695399/functions/posix_openpt.html
