* Update network user guide to point to ansible.utils.cli_parse module as ansible.netcommon.cli_parse is deprecated from ansible.netcommon 2.0.0 release
* modules moved to use best_parsable_locale
* fixed invocations
* better better
* also module_utils
* converted to function as per fb
* patch testt
* whitespace
* set host_key_checking check to False, rather than if not (which captures False and None)
* add host_key_checking default to ssh.py / update documentation
* Canonicalize module_defaults actions and action_groups pre-fork and cache them on the play
* Call get_action_args_with_defaults with the resolved FQCN plugin and don't pass the redirect list
* Add validation for action_group metadata and a toggle to disable the warnings
* Handle groups recursively referring to each other
* Remove special-casing for non-fqcn actions in module_defaults groups
* Error for actions and groups in module_defaults that can't be resolved
* Error for fully templated module_defaults
* Add integration tests for action_groups
* Changelog
This is useful for testing the effects on docs of changing what
collections are in a version of ansible. ansible-build-data can be
cloned locally, modified, and then build-ansible.py docs-build can be
told to use the local checkout instead of the canonical one on github.
* env_setup - Prefer Python3
Prefer the 'python3' executable over 'python'.
* Add shebang for syntax detection
While the script isn't executable and we recommend sourcing it, having a shebang allows
variuos text editors to detect that it is a shell script and apply proper syntax highlighting.
While implementing basic Ansible module testing I realised that the `self.module` used in some places is not actually a thing, and I changed to `my_module` to make it work as I understand it was expected.
Since the task and connection both have the same 'retries' keyword, the task default
would override the connection value.
Do not pass 'retries' from the task to the connection options.
* Set ssh_connection retries default value back to 0
It was 0 before the move to config and was changed to 3 by accident.
This version relies on an external lib implementing lexers and styles.
That lib pulls in proper Pygments version so this patch also drops it
from direct requirements.
This is a follow-up of previous years' 1a11cec and c8315bf. It deals
with links which at that point presumably either were not present or
did not support https://
* module_common - handle cache directory creation collision
Occasionally multiple workers can try to create the cache directory if it exists. Catch the exception
and handle it.
* Just re-raise if it doesn't exist rather than trying again
* Use exists_ok rather than handling an exception
* Remove unused import and unused variable
* Go back to try/except but with OSError
Since we do not have split controller remote, this needs to be able to run in Python 2 in order
to be merged currently.
Since we moved yum_repository_test_repo.baseurl to file:/// format,
using urlsplit filter on it returned an empty string. Using an empty
string as a left operand of `in` was then always evaluated to True.
This was discovered with native jinja being on where urlsplit returns
None which results in a hard fail as None cannot be a left
operand of `in <string>`.
* The resolved_action is the formatted version of the final plugin in the PluginLoadContext's redirect_list
* Collection plugins are represented as FQCN
* Legacy plugins are represented with only the plugin name
* Add tests
* Changelog
* Fix struture of cli/doc.py snippet code.
A couple releases ago, cli/doc.py was modified to mostly conform to the
data processing pipeline steps. format_plugin_doc() was the biggest
exception in that refactor. When the snippet code was made generic
instead of being only for modules, the new code should have conformed to
the data processing pipeline too.
* Move the decision to output a snippet to the run() method alongside
the decision to output a listing versus plugin_docs.
* Move the test for invalid plugin_types to the run() method as it
affects all snippets in this run, not just a single snippet. (-t can
only be specified once)
* Rename get_snippet_text() to format_snippet() as:
* This is the data formatting step
* The format_snippet() name matches with its conceptual sibling,
format_plugin_doc().
* Use ValueError inside of format_snippet() to flag unrecoverable errors
formatting a single snippet.
* Emit a warning when format_snippet() raises ValueError and continue to
the next snippet.
* If the yaml(?) or toml inventory plugin is specified for snippet output,
raise ValueError() so that the user sees a warning instead of simply
seeing blank output.
* Do not modify arguments passed into format_snippet(). This is the
formatting step so data should not be modified.
* Change _do_yaml_snippet() and _do_lookup_snippet() to operate side
effect free.
* Fix raising of exceptions when formatting requred options for snippets.
* Unrelated: Use to_text() instead of to_native when calling
display.warning(). to_native() is used for raising exceptions. Not
for display methods.
* Add a changelog
specially for when you have parameters in unicode but need
to scrape responses, C is still the fallback
Co-authored-by: Abhijeet Kasurde <akasurde@redhat.com>
* Require antsibull 0.34.0.
* Remove Makefile comment about the devel docs building only the core
(formerly base) docs; the behavior was updated in ccbfdec334
Currently 'make coredocs' builds core-only docs; 'make webdocs' builds the full docs even on the devel
branch
Co-authored-by: Toshio Kuratomi <a.badger@gmail.com>
* removes upper bound on sphinx version
* updates versions of docs build dependencies, adds known good requirements file
* adds instructions for using known_good_reqs file
Can now specify plugin/plugin type for list and dump actions
New 'init' action to create usable config sources
Co-authored-by: Felix Fontein <felix@fontein.de>
Alicia Cozine