Now, "bup fuse dir", assuming ~/.bup is one's bup repository, will
create a fuse mount on dir, in which one can see the backups (with
lots of rought edges).
ac_cv_file__proc_meminfo=yes to "configure" script. As a result htop
will be built regardless of /proc is mounted or not (bulk builds).
Enable htop PR 39881 on NetBSD, Linux and FreeBSD only where
Linux-compatible procfs is available.
An implementation of several specifications from freedesktop.org intended for
use in Enlightenment DR17 (e17) and other applications using the Enlightenment
Foundation Libraries (EFL).
Currently, the following specifications are included:
o Base Directory
o Desktop Entry
o Icon Theme
o Menu
Upstream changes:
2011-11-23: released Fabric 1.2.5
2011-11-23: released Fabric 1.1.7
[Bug] #441: Specifying a task module as a task on the command line no
longer blows up but presents the usual "no task by that name" error
message instead. Thanks to Mitchell Hashimoto for the catch.
[Bug] #475: Allow escaping of equals signs in per-task args/kwargs.
[Bug] #450: Improve traceback display when handling "ImportError"s
for dependencies. Thanks to David Wolever for the patches.
[Bug] #446: Add QNX to list of secondary-case sed targets. Thanks to
Rodrigo Madruga for the tip.
[Bug] #443: exists didn't expand tildes; now it does. Thanks to
Riccardo Magliocchetti for the patch.
[Bug] #437: with_settings now correctly preserves the wrapped
function's docstring and other attributes. Thanks to Eric Buckley for the
catch and Luke Plant for the patch.
[Bug] #400: Handle corner case of systems where pwd.getpwuid raises KeyError
for the user's UID instead of returning a valid string. Thanks to Dougal
Matthews for the catch.
[Bug] #397: Some poorly behaved objects in third party modules triggered
exceptions during Fabric's "classic or new-style task?" test. A fix has
been added which tries to work around these.
[Bug] #341: append incorrectly failed to detect that the line(s) given
already existed in files hidden to the remote user, and continued appending
every time it ran. This has been fixed. Thanks to Dominique Peretti for the
catch and Martin Vilcans for the patch.
[Bug] #342: Combining cd with put and its use_sudo keyword caused an
unrecoverable error. This has been fixed. Thanks to Egor M for the report.
[Bug] #482: Parallel mode should imply linewise output; omission of this
behavior was an oversight.
[Bug] #230: Fix regression re: combo of no fabfile & arbitrary command use.
Thanks to Ali Saifee for the catch.
Changes since version 1.0.30:
- A race condition in key generation has been fixed which could
allow a newly-generated key file to be read by another local
user if the key file is being generated in a world-readable
directory and the user running tarsnap-keygen has a umask other
than 0066.
- A bug in key generation has been fixed which could allow a
newly-generated key file to be read by another local user
if they key file is being generated in a world-writable
directory (e.g., /tmp).
- Tarsnap now supports Minix.
- Tarsnap now ignores blank lines in key files; line-buffers
its output (which makes tarsnap --list-archives | foo more
responsive); and prints a progress indicator during tarsnap --fsck.
- Multiple minor bug fixes.
mrsync is a tool to transfer files from a master to many remote
machines in LAN using Unix multicast sockets. It dynamically adjusts
its transfer speed to ease on the network and to leave no one behind.
4 hours for 140GB to 100 targets in 1Gbit LAN
While converting libgtop to use SPECIAL_PERMS, too much of the
previous patch to the configure script was removed, breaking DragonFly.
Restored the missing bits to unbreak DragonFly.
Upstream changes:
2011-11-07: released Fabric 1.2.4
[Support] #459: Update our setup.py files to note that PyCrypto released
2.4.1, which fixes the setuptools problems.
[Support] #467: (also #468, #469) Handful of documentation clarification
weaks. Thanks to Paul Hoffman for the patches.
per maintainer update request by PR pkg/45630.
Changes:
* Corrected compilation when --with-ipa-pthread or
--with-ipastat-pthread was specified.
* Corrected format of tables in "ipactl memory" output.
* Corrected access to free memory in ipa when dynamic rule is deleted.
* Call log_stderr() and log_stdout() after ipa.conf file parsing.
* Simplified and renamed confcommon.c:print_spaces().
* Corrected output of "ipa -v" and "ipa -h".
Don't complain if there is no custom-kernel file, and don't try to
build GENERIC as the default custom kernel, because it's built
anyway if appropriate for that port.
Support .branch to define branch for builds from NetBSD imported
into a VCS, rather than checked out from NetBSD's CVS. Example
contents for .branch is "5\n" for netbsd-5.
Support local and relative paths in .tree, for placing the obj
container in arbitrary locations.
This package creates ${LOCALBASE}/share/applications directory, but it
didn't attempt to remove it upon deinstallation. DEINSTALL fragment has
been updated to fix this.
Changelog:
## 2.9.0 / September 24 2011
A vairly heavy release, including some new features which most of you won't
need, but power users have contributed, this also marks the beginning of the
end of the 2.x series, more information will follow in due course, but with
the proliferation of Bundler, and better ways to do deployment, we will be
introducing heavier changes to Capistrano to keep the tool current.
**Please note, following some reported problems with the asset pipeline code
being not found, remember Capistrano needs to be in your Gemfile, and as such
needs to be run with Bundler, otherwise you risk loading a system-wide version
of Capistrano who's behaviour might be different from that specified in your
Gemfile. This is also good practice because much of the deploy logic resides
in the Gem, and you wouldn't want that to change without your knowledge. Rails
applications include Cap in the Gemfile anyway, you should follow this
convention.**
* find_servers() will no longer raise if there are no servers, this behaviour
can be modified by way of the `:on_no_matching_servers` option. Thanks to
`@ppgengler`.
* Short Git SHA1 fragments are now supported in commands such as `cap deploy
-s revision=d7e99f` thanks to `@ndbroadbent`.
* One can now specify individual SCM commands by setting
`:scm_arguments_<command_name>`. Thanks to `@alextk`.
* Travis CI build now passes thanks to @andrew, build tested against MRI
`1.8.7`. `1.9.2` and `REE`.
* Support for Perforce labels, I don't know much about this, but I believe
it's much like deploying a Git tag, thanks to `@ak47`.
* Git SCM now correctly adheres to the `:scm_verbose` setting. Thanks
`@dubek`.
* `set()` can now be used to set a `false` value, previously this was a no-op.
Thanks to `@nilbus`.
* Support for Git 1.6x submodules, The Git SCM strategy now queries Git on the
server-side to ensure it supports the `--recursive` flag, if it doesn't then
it will fall back to using the long-hand. Many thanks to all those involved in
the discussion surrounding this topic, and to `@nilbus` for a beautifully
clean solution which doesn't hold us back.
* When using `:cached_copy` with Subversion, use `svn switch` to for more
reliable switching of branches/etc. Thanks to `@iGEL` for the patch that we
accepted finally, and to `@richmeyers` who also submitted a patch and
contributed to the discssion.
Other cleanups and minor improvements to the code and tests were committed by yours truly
(@leehambley), @maxim, @ak47 and @andrew).
