This makes the following changes:
- The `append_newline` argument to the `file.blockreplace`
remote-execution function has been modified so that if its value is
`None`, it only appends a newline when the content block does not end
in one.
- A couple of fixes were made to newline handling. The existing code
normalized the newlines in the content block, replacing them with
os.linesep. However, when the file contains newlines that don't match
the OS (i.e. POSIX newlines in a file on a Windows box, or Windows
newlines on a Linux/Mac/BSD/etc. box), then we would still end up with
mixed newlines. The line separator is now detected when we read in the
original file, and the detected line separator is used when writing
the content block. Additionally, the same newline mismatch was
possible when appending/prepending the content block. This has been
fixed by using a common function for appending, prepending, and
replacing the content block.
- Support for the `append_newline` argument has been added to the
`file.blockreplace` state. The default value for the state is `None`.
A `versionchanged` has been added to the remote execution function to
let users know that the Fluorine release will change the default value
of that variable.
- 20 new integration tests have been written to test the
`file.blockreplace` state.
If the ssh binary is not found in path, SSHPasswordTests will fail:
ERROR: test_password_failure (unit.ssh.test_ssh.SSHPasswordTests)
[CPU:0.0%|MEM:54.9%]
----------------------------------------------------------------------
Traceback (most recent call last):
File "tests/unit/ssh/test_ssh.py", line 42, in test_password_failure
client = ssh.SSH(opts)
File "salt/client/ssh/__init__.py", line 226, in __init__
raise salt.exceptions.SaltSystemExit('No ssh binary found in path -- ssh must be '
salt.exceptions.SaltSystemExit: None
Either SSHPasswordTests needs to be converted into a real unit test
(i.e. mocking the salt call) or this tests needs to be skipped if ssh is
not available.
Signed-off-by: Benjamin Drung <benjamin.drung@profitbricks.com>
There is a note about deprecating cmd.wait in the docs of the cmd state.
However, this function has not been put on an official deprecation path
and there are no plans to deprecate it. Therefore, we shouldn't have
this note in the docs.
See #44831 for more information.
We had a similar check to ensure "foo: bar" isn't loaded as a dict, this
adds a check to ensure that we only load lists when there is a leading
bracket.
In some edge cases, the branch/directory paths used here could have
values loaded as int/float/etc., which is not a problem for
salt.utils.gitfs since it normalizes these values to strings. Therefore,
this commit removes them from VALID_OPTS so that we skip type-checking
to avoid spurious warnings in the log file.
