| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| ... | |
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This splits out the plumbing into the Puppet::Interface namespace, and uses
Puppet::Faces for all the public-facing code.
The fault line is "what you care about if you are using or writing a face",
which is public, against "what you care about to enable either of those two",
which is the plumbing.
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
The codebase is now using the new name, faces, uniformly to reference the
objects contained. All tests pass.
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
This just changes filenames and directories; files are exact copies rather
than having additional modifications to make clearer each step of this
process.
This does leave a currently broken build. :/
|
| | | |\ \ \ \ \ \ \ |
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
This was causing failure in some cases, based on load order; we should always
satisfy our external dependencies.
Reviewed-By: Dan Bode <dan@puppetlabs.com>
|
| | | |\ \ \ \ \ \ \
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
'bug/master/6972-setting-CA-location-for-cert-string-no-longer-works'
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
The certificate string is broken, and won't allow you to either search or save
certificates. Given that, mark the test on it broken until that is completed.
Reviewed-By: Dan Bode <dan@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This ports the existing certificate location configuration to be a string
option, and then uses that to change the configuration. This will leak state
between calls, which is somewhat unavoidable, but should at least get the
basic stuff right for the CLI.
We eventually need the CA string to be supported by a stateless internal CA
implementation that allows us to do the right thing overall.
Reviewed-By: Dan Bode <dan@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This extends the CLI pre-parse phase to identify both string *and* global
options out of the Puppet settings/defaults system. This makes the regular
CLI support for setting Puppet configuration globals work as expected.
This moves us along the line of supporting these options more fully.
Reviewed-By: Dan Bode <dan@puppetlabs.com>
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
During testing, an obvious cleanup showed up for the name extraction here, so
we implement it. This extends the regexp to better extract the data we want
rather than hacking it up post-match and having to do extra validation to make
sure it actually worked.
Reviewed-By: Dan Bode <dan@puppetlabs.com>
|
| | | |\ \ \ \ \ \ \ |
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We do this by implementing a standard mechanism for finding the current
version out of the default file, and only supporting that one file. This
implements our decision to lazy-evaluate the extra version support stuff as
much as possible.
Reviewed-By: Dan Bode <dan@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We had a pattern where we wanted to routinely write scratch strings to disk,
on the load path, so we could reference them later. This extracts that into a
helper, and starts to use it in tests that should follow that pattern.
Reviewed-By: Dan Bode <dan@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We had some incorrect indentation in the file; we fix that up now before we
make any substantive changes.
Paired-With: Dan Bode <dan@puppetlabs.com>
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
We were flushing loaded string versions in the test suite, but were not
clearing out the $" array used to determine if we had previously loaded that
file.
This lead to situations where tests cross-polluted because require though the
code was in memory, but the effects of loading that code had all been rolled
back.
Reviewed-By: Dan Bode <dan@puppetlabs.com>
|
| | | |\ \ \ \ \ \ \ |
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
Reviewed-By: Matt Robinson
|
| | | |\ \ \ \ \ \ \ |
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
We had a logic failure that didn't pass positional arguments at all, but which
our testing didn't verify. This entirely broke things. Now fixed, and a test
added to ensure we don't bug out further...
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | |\ \ \ \ \ \ \ |
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
In the DSL we want to use 'when_invoked do' because it reads much more
naturally for users.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
|
| | | |\ \ \ \ \ \ \ |
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We didn't correctly handle '--foo=bar' as having supplied an argument during
the pre-parse phase. Now we have a test for it, and a fix in the code.
Reviewed-By: Nick Lewis <nick@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This adds a test to verify that we are correctly removing the action name from
the set of arguments passed to the string action, then cleans up the previous
code so we don't need to mutilate the command line arguments: we can just
extract it from the resultant set of information.
Reviewed-By: Nick Lewis <nick@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
optparse will treat '--foo --bar' as "foo with the argument --bar" when foo
takes a mandatory argument. We need to emulate that behaviour in our
pre-parse of the command line.
Incidentally, fix up a bug in boolean options, and improve our testing.
Reviewed-By: Nick Lewis <nick@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This improves handling of the pre-parse of the command line to be
non-destructive, which cuts down the volume of garbage generated in the
process.
It also improves testing to verify that we get the darn thing right...
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This cleans up a whole bunch of bits of the testing code around the place;
nothing revolutionary, just nicer and more robust code.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We now accept a terminus option to each invocation, and set the terminus based
on that call. This is probably incomplete, because it only sets the terminus
when given, and doesn't try to reset it to the default afterwards.
This also resets the terminus class after every invocation, to stop it leaking
state across calls. This make, sadly, have some effects if you are not just
using the strings to invoke the terminus, but it beats having the strings
broken as well...
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
To present a pleasant Ruby API to folks invoking actions we want to have
default values for the trailing 'options' argument, and to be able to pass a
block to the code to allow yield to work.
Given limitations of Ruby 1.8 regarding blocks, we can't use either of those
because the block slot is bound at declaration time, and you can't give
optional arguments.
To work around this we inject, using eval, a full regular Ruby method into the
final block of code. This can provide the default argument at the end in an
intelligent way (albeit by emulating what the interpreter would do).
This doesn't solve the yield problem, but the same method can pass the block
explicitly, and allows other hooks to be injected, which makes it easier to do
smarter things in future...
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This provides a solid test of the new code, by migrating the existing strings
to match. This also gives us a chance to determine any weak points in the
code as written.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This removes dead code now we have terminus in the base string, and disables
some tests on StringBase app until they can be rewritten.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We used to generate an info-level message, then carry on, which typically
resulted in raising because 'nil' doesn't implement the expected method that
we immediately try to call.
So, instead, raise a clear error at the time we fail to load, which gives a
pleasant rather than confusing error to the user. Which at least means they
know why things have gone wrong...
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
So, we have Action#invoke, but it binds to the declaring class, not to the
correct instance. Solving all the subtle issues around threads, global state,
and bindings without causing us too much pain is actually pretty hard, so
instead we pull the feature.
It can be enabled again in a future release and, being a strict extension
feature, we can do that without overly hurting anyone. We still have full
access to the invocation through a marginally less pleasant syntax, but one
that people MUST be able to arrange to allow invoke to work, so on that front
we are clean.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
Now we are pushing into production we can eliminate this language, which was a
legacy from the prototype that is no longer relevant globally.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
In order to identify the full set of options we need to know the action that
is being invoked; that actually requires a pre-processing step to identify
that out of the global options.
Notably, our spec is that options can be to the right of their declaration
point, but not to the left: that means that we can now extract the full set of
options, and interact with the standard Puppet option handling code, resulting
in at least vaguely saner behaviour...
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This ensures that an option declaration that shadows itself is found, and
reported to the user, rather than silently eating one of the two.
This could have actually lost, for example, the distinction between an
argument-requiring and an argument-missing variant of the same thing.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
When we wrote class inheritance of actions for strings we didn't implement
method (ahem, action) lookup correctly. This changes that, by providing the
implementation to our standards, along with appropriate tests.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
The purpose of this is to adapt the generic option support in our strings to
the command line; we adapt the generic option information to optparse, and
establish our environment early in the process to ensure that we can play nice
with Puppet::Application for the moment.
In the process we ensure that we detect, and report, conflicts in option
naming across the board. Additionally, when an option is declared with
multiple aliases, we insist that either all, or none, of them take an
argument.
To support this we support introspecting options having an optional argument,
as well as documentation and all.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We want to support both strings and actions specifying options, to support
generic wrappers that present strings to the user across multiple distinct
front-ends.
At the moment we focus on implementation of a generic CLI providing full
control to all the strings, but we aim to support other programmatic
interfaces including Ruby and RPC invocation as part of the overall change.
We also detect, at the time they are declared, duplicate options. They are
reported, like any duplicate, with an error thrown. Specifically:
It is illegal to declare a duplicate option in the same scope, such as within
the same string, or within the same action. This is unchanged.
It is illegal to declare an option in an action that duplicates an option in
the string, or vice-versa. This is reported when the duplicate is declared,
so may report on either the string or action depending on sequence.
It remains legal to duplicate the same option across multiple actions, with
different meanings. There is no conflict, as the same option can't be passed
to both simultaneously.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
This makes it possible for us to just print the Action object directly and get
a human-focused output string.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
Earlier in their implementation the String prototype would set global state on
a String object to reflect options set on the command line. As we move
strings away from a CLI-only prototype, this becomes troublesome because we
can easily have, for example, HTTP access to a string, which means load
balancers can really make this confusing.
It also encourages global state pollution, where one invocation can adversely
influence another. A better approach is that we pass options to the string
action invocation directly; this makes the interaction stateless.
Changes required to your code to adapt to the new world:
- action(:foo) do |some, args|
+ action(:foo) do |some, args, options={}|
if options[:whatever] then
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
Previously we tried to invoke a class method on an instance, and threw another
exception, which wasn't the most helpful behaviour. This fixes that to
correctly report the array of available terminus classes to the user.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
The methods being tested are not available in the 2.6 series, and the test
itself is relatively weak, so rather than try to fix it we eliminate it in
favour of other testing of the same behaviour.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We had an outstanding pending test for code we wrote, and which we were not
actually testing stand-alone. This implements the test for that.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | |
| | | | | | | | | | |
We had a series of tests that vary only on one input, and which wanted to
validate that the same failure happened after every instance; reducing this
to a set of input data lets us reduce code duplication in the tests.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
We depend on require throwing about a specific file on disk, rather than just
generically, to ensure that we don't consume load errors from existing files.
This means our tests need to be updated to reflect that, by raising with
appropriate text when they stub require.
Reviewed-By: Pieter van de Bruggen <pieter@puppetlabs.com>
|
| | | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
Paired-With: Matt Robinson
|
| | | |\ \ \ \ \ \ \ |
|
| | | |/ / / / / / /
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | |
| | | | | | | | | |
The intent of these specs is to find the String rather than to actually define
it. Thus, the bracket notation is more semantically accurate than using
Puppet::String#define.
Reviewed-By: Pieter van de Bruggen
|
