Luke Kanies wrote:
> On Sep 15, 2008, at 3:29 AM, Brice Figureau wrote:
>
>> Brice Figureau wrote:
>>> Hi,
>>>
>>> Here is an attempt to provide an append variable operator (see below
>>> for an example). My minimal testing show that it seems to work fine
>>> in simple manifests.
>>>
>>> My aim is to gather comments/reactions from the community regarding
>>> this feature which was previously asked (and that I wanted myself).
>>> Let me know if that's something that _shouldn't_ be done: I know
>>> puppet vars are write-only by design, so changing their values might
>>> open a pandora box I'm not aware of.
>> I meant write-once of course, and not write-only.
>> Note that it is possible to allow the append operator only in a new
>> scope and raise and error otherwise.
>> Ie allow sth like:
>>
>> $var = 'puppet'
>> class klass {
>> $var += ' is great'
>> }
>>
>> but not:
>> $var = 'puppet'
>> $var += ' is great'
>>
>> It is just a matter of checking that the scope doesn't yet contain the
>> variable.
>
>
> I'm not opposed to this, and can see why people are interested in this.
>
> Just to be clear, this works by basically making a new variable in the
> current scope that is a concatenation of the specified variable and
> the new text/array, right?
Not a new variable, it adds in the current scope a variable with the
same name as the upper scope variable but whose value is concatenated
with the current rvalue.
The implementation in the patch above allows to append _in the same
scope_, but I don't think this is right, hence my aforementioned comment
that I can easily explicitely allow append operator only on new
descendant context.
> If so, then that seems perfectly reasonable to me, since it's very
> similar to just using something like:
>
> $othervar = "$var new text"
>
> It only affects lower scopes, rather than modifying the original
> value, so it should work great.
Yes.
> As to tests, there should be sufficient lexer tests, but I agree that
> we don't have parser tests to speak of; all I've got are integration
> tests in the form of snippets, like you already have done. It might
> be a good time to create that parser unit test file and add tests for
> this code. :)
My problem is that I don't even know how to start a parser unit rspec :-(
The only thing I can think of is: parse some puppet snippets (or should
I feed lexer results?), then mock the AST::XYZ object to check that they
are called fine. Is that the right thing to do?
Right now, I have a rspec append test, that is more an integration test,
as it feeds puppet snippets to a parser, then compile the AST tree, then
finally checks that the scope contains the right value, or that right
scope function setvar is called. Is that something you might include?
> I think the major changes, though, are in the scope class, and it's
> also only tested in test/ rather than spec/.
OK, so I can also start a rspec scope that at least check the setvar method.
On a different side, there are demand for more complex if test (ie
including or, and, ==, !=, etc ala collection expressions). I'm
volunteer to implement this (and even already started). Is this
something that you can include, or should I use my spare time for
something else?
Thanks,
--
Brice Figureau
Days of Wonder
http://www.daysofwonder.com
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Puppet Developers" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/puppet-dev?hl=en
-~----------~----~----~----~------~----~------~--~---
