| Commit message (Collapse) | Author | Age | Files | Lines |
|\
| |
| | |
Removing ole_initialize/uninitialize
|
| |
| |
| |
| | |
This seems to stop ruby from seg faulting on Windows
|
| |
| |
| |
| |
| | |
The doc changes were removed from the last release, and so not present
in this commit.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The following code is brittle if you want to fork the cookbook and rename it
(which is a best practice when forking cookbooks for internal use):
```ruby
if node[:platform_family] == "rhel"
include_recipe "mycookbook::_rhel"
end
```
In order for cookbooks to be easily renamable they can currently use the
syntax `include_recipe "#{cookbook_name}::_rhel"` which is unwieldy.
This patch adds `include_recipe "::_rhel"` as syntax sugar to make this
easier.
|
| | |
|
| | |
|
|\ \
| | |
| | | |
deep_merge_cache fixes for bugs in 12.0.0
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
In 12.0.0 we introduced a cache for the merged attributes for the
top-level node attribute keys. This fixes this so that node['foo']
and node[:foo] are not cached separately. This also showed up in bugs
as issues between node['foo'] access and node.foo access because
node.foo is translated into node[:foo].
|
|\ \ \
| | | |
| | | | |
bugfix dscl provider
|
| | | | |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
We need to compare with the UID, not a GUID. So we need to map the
correct field for comparison.
Also fix a bug in determining used uids.
This fixes this exception when the specified uid is actually in use by the
correct user:
Chef::Exceptions::RequestedUIDUnavailable: uid 48 is already in use
|
|\ \ \ \
| | | | |
| | | | | |
OS X user provider - fix exception if no salt is found
|
| | | | | |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Some system users don't have salts. This exception is fixed by this
commit:
# in `convert_to_binary':
NoMethodError: undefined method `size' for nil:NilClass
|
|\ \ \ \ \
| |_|_|/ /
|/| | | | |
Macports provider - provide package
|
| | |/ /
| |/| |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The MacPorts provider should provide package so that you can change the
precedence in provider mappings, with a statement like this:
Chef::Platform::ProviderPriorityMap.instance.priority(
:package, Chef::Provider::Package::Macports, :os => 'darwin'
)
Also, set default priority in mapping so that homebrew wins.
|
| |/ /
|/| | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Turns the 3694 warning into a debug message if the prior resource is
identical to the current resource. Suggestion for opscode/chef-rfc#76
This also moves resource building out to its own class.
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
the default is still the same, but will warn that users should start
being explicit. setting compile_time false will become the new
default in chef-13 and is encouraged to avoid the compile_time arms
race.
|
|/ /
| |
| |
| |
| |
| | |
if we don't force LANGUAGE then package installs will still fail, etc
due to taking precedence over even LC_ALL, force LANG as well for
good measure which should cover all the bases hopefully.
|
| | |
|
| |
| |
| |
| |
| | |
macports needs to provide macports_package to `darwin`, not `mac_os_x`.
`mac_os_x` is the platform, `darwin` is the OS.
|
| | |
|
|\ \ |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
manner.
This is as per http://en.wikipedia.org/wiki/URI_scheme, and solves some
edges i.e., following (30x) URL from the "Location" header where we
have to deal with "HTTP://".
Signed-off-by: Krzysztof Wilczynski <krzysztof.wilczynski@linux.com>
|
| |\ \
| | | |
| | | | |
Make search with filtering match partial_search.
|
| | | | |
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
* Exclude rows from query if rows are nil.
* Update defaults in knife/search to match query defaults.
* Update start when polling more rows.
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | | |
|
| | | |
| | | |
| | | |
| | | | |
* Updated search to use argument parameters.
|
| |\ \ \
| | |/ /
| |/| | |
Enable logon-as-service in windows_service (CHEF-4921).
|
| | |/
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
:run_as_user and
:run_as_password attributes to the windows_service resource. If a logon user is specified,
the resource will (on every run) grant the logon-as-service privilege to that user, using
secedit.exe.
|
|\ \ \
| |/ / |
|
| |\ \
| | |/
| |/| |
Fix bug where errored parsing from what-if output causes resource to be considered converged
|
| | | |
|
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
converged.
We now fall back to assuming the resource is not converged if we cannot parse
information about a dsc resource.
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | |
| | | |
* Audits are disabled by default.
* Also, updated spec file to use RSpec :let.
|
| | | |
|
| | |
| | |
| | |
| | |
| | | |
* Modify command line option --audit-mode to accept parameters enabled, disabled, or audit-only.
* Emit a warning if audit-mode is enabled or audit-only.
|
| | | |
|
| | | |
|
| | | |
|