On May 31, 2009, at 6:21 AM, Brice Figureau wrote:
>
> On 31/05/09 1:15, Luke Kanies wrote:
>> On May 30, 2009, at 1:17 PM, Brice Figureau wrote:
>>
>>> I don't know why we imposed the restriction that we shouldn't match
>>> with parameter containing arrays in exported mode.
>>> That doesn't seem right, as the produced rails query works fine with
>>> arrays.
>>
>> I just noticed, too, that the current code doesn't allow conditionals
>> in the Rails query, which also shouldn't be right.
>>
>> I'm hesitant to fix it at this point, though, since we still need to
>> move our Rails query to a more abstract, pluggable query (e.g.,
>> something through a RESTful interface), so keeping it simple until
>> that's done is a good idea.
>
> It's simple. It's just matching multi-valued parameters.
> I think I didn't explain correctly what the problem is.
>
> The in-memory query forbids matchin multi-valued parameters, but the
> rails one allows it by design.
> Ex:
>
> @@file {
> ..., tag => ["a","b"]
> }
>
> File<<| tag == "a" |>>
>
> This won't match for the host who created the resource when it
> collects
> it, but will work for other hosts, because the rails query is
> (pseudo sql):
>
> param_name.value="tag" and param_value.value="a"
>
> which works fine for multi-valued parameters.
I see.
>
>>> Note: the user tags are not stored in the rails database except
>>> under
>>> the special resource parameter tag. This also doesn't seem right.
>>
>> Really? So if someone sets 'tags' parameters on a resource, they're
>> not stored in the db as parameters?
>
> No they're stored as a parameter but not as a tag.
> And the built rails query matches tag as parameter.
> So you can't do:
>
> class myfile {
> @@file { ... }
> }
> File<<| tag == "myfile" |>>
>
> My idea would be that we should persist user tag as tag _and_
> parameter
> and that tag queries should query the persisted tags.
I agree -- the two tag lists should be entirely equivalent. They are
on the client, but since we rely on tags on the server, they should be
made equivalent there, too.
>
>>> Signed-off-by: Brice Figureau <brice-pup...@daysofwonder.com>
>>> ---
>>> lib/puppet/parser/ast/collexpr.rb | 2 +-
>>> spec/unit/parser/ast/collexpr.rb | 8 +++++---
>>> 2 files changed, 6 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/lib/puppet/parser/ast/collexpr.rb b/lib/puppet/parser/
>>> ast/collexpr.rb
>>> index 6ade58b..52f930e 100644
>>> --- a/lib/puppet/parser/ast/collexpr.rb
>>> +++ b/lib/puppet/parser/ast/collexpr.rb
>>> @@ -31,7 +31,7 @@ class CollExpr < AST::Branch
>>> when "and"; code1.call(resource) and code2.call(resource)
>>> when "or"; code1.call(resource) or code2.call(resource)
>>> when "=="
>>> - if resource[str1].is_a?(Array) && form != :exported
>>> + if resource[str1].is_a?(Array)
>>> resource[str1].include?(str2)
>>> else
>>> resource[str1] == str2
>>
>> I don't see how this code could fix things - this is only for the
>> query against in-memory resources, not in the database, right? So
>> you'd match those resources from the current host, but not from all
>> hosts.
>
> Correct. The other host case was already covered by the rails query
> produced.
>
>> I think there needs to be a special case for tags in the Rails query,
>> just like there is for title.
>
> Yes.
>
>> And there should probably be something on the export side that
>> unifies
>> the tags set as parameters (e.g., tags => foo) and those set on the
>> resource itself through parsing.
>
> Correct, too, hence my side not in my original commit msg.
> So what's the plan?
> * leave it as is?
> or
> * apply this patch?
> or
> * enhance the patch to unify tags (coming for the tag parameter, the
> tag function, the scope) in tag db and adapt the rails query?
>
> Choose your poison :-)
I prefer to unify the tags. I'm a touch uncomfortable about it, since
we try to keep the parser from knowing much about resource attributes,
but there's not much we can do about it here.
--
It is odd, but on the infrequent occasions when I have been called upon
in a formal place to play the bongo drums, the introducer never seems
to find it necessary to mention that I also do theoretical physics.
-- Richard Feynman
---------------------------------------------------------------------
Luke Kanies | http://reductivelabs.com | http://madstop.com
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Puppet Developers" group.
To post to this group, send email to puppet-dev@googlegroups.com
To unsubscribe from this group, send email to
puppet-dev+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/puppet-dev?hl=en
-~----------~----~----~----~------~----~------~--~---
