I think this hits the nail on the head.

post_resources_eval ==> after all the resources of this type
post_resource_eval ==> after this resource

And, yes, we need both as well as post_catalog_eval!

I'd personnal prefer that there is not an explicit failure state between
these but, instead, state data that is passed between them.

post_resource_eval -> Hash(success_status) -> post_resources_eval ->
Hash(success_status) -> post_catalog_eval.

This way you can do different things based on what happens as a controlled
cascading failure without digging through the entire catalog.

Also, it would be nice for the failure state to be part of each run (and
reset) in such a way that dependent resources (autorequires, etc...) can
make a decision based on the state of what it's depending on.

Trevor


On Fri, Mar 7, 2014 at 3:09 PM, James Sweeny <[email protected]>wrote:

>
>
>
> On Fri, Mar 7, 2014 at 1:16 PM, Jeff McCune <[email protected]> wrote:
>
>> On Fri, Mar 7, 2014 at 7:45 AM, Nan Liu <[email protected]> wrote:
>>
>>> On Thu, Mar 6, 2014 at 10:16 PM, Jeff McCune <[email protected]>wrote:
>>>
>>>> This is just a continuation of a previous thread as to not hijack the
>>>> original discussion.
>>>>
>>>> The question that needs a decision is, should post_resource_eval be
>>>> renamed given the context that it's currently implemented as a hook into
>>>> the point after all resources for a provider are evaluated and we might
>>>> want a hook into the point after each discrete resource is evaluated?
>>>>
>>>> Nan agrees it should be renamed hence the need for a decision.
>>>>
>>>
>>> I'm not sure where this should be documented (is there a ticket?). There
>>> is one more challenge for post_resource_eval v.s. post_catalog_eval. I
>>> believe there is a need to be able to establish a dependency to
>>> post_resource_eval.
>>>
>>
>>
>> Yeah, I'm thinking the idea of "post catalog eval" is flawed because it's
>> really not after the _catalog_ but rather all provider instances for a
>> given resource type.  There may still be quite a bit of the catalog left to
>> evaluate when the hook fires.
>>
>> post_type_providers_eval maybe?
>>
>>
>>
> post_resources_eval as opposed to post_resource_eval? Is that too small a
> syntactic distinction?
> --
>
> James Sweeny
> Professional Services
> http://puppetlabs.com/
>
> *Join us at PuppetConf 2014, September 23-24 in San Francisco - *
> http://bit.ly/pupconf14
>
>  --
> You received this message because you are subscribed to the Google Groups
> "Puppet Developers" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/puppet-dev/CAKDACKu0%2BaJnrVO-N8P4mgKvTQ08hrABdsR6wQXCyifobsnrRQ%40mail.gmail.com<https://groups.google.com/d/msgid/puppet-dev/CAKDACKu0%2BaJnrVO-N8P4mgKvTQ08hrABdsR6wQXCyifobsnrRQ%40mail.gmail.com?utm_medium=email&utm_source=footer>
> .
>
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Trevor Vaughan
Vice President, Onyx Point, Inc
(410) 541-6699
[email protected]

-- This account not approved for unencrypted proprietary information --

-- 
You received this message because you are subscribed to the Google Groups 
"Puppet Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/puppet-dev/CANs%2BFoVS037yAOj5FNNVybxhQQ7p_WnAmj5Rs3dDVr8fKS5pBA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to