1) Yes, the two capability names are in the same node type, the name
distinguishes them. (This is a YAML issue.)
2) Yes, the two are unrelated and can be the same or different.

On Thu, Oct 26, 2017 at 1:58 PM, Steve Baillargeon <
[email protected]> wrote:

> Let me use a concrete example by modifying the compute node type a little.
> See below.
> Assume this is for a single node type
>
> ....
> requirements:
>   - <requirement definition name 1>:
>       capability: tosca.capabilities.Attachment
>       node: tosca.nodes.BlockStorage
>       relationship: tosca.relationships.AttachesTo
>       occurrences: [0, UNBOUNDED]
> capabilities:
>   <capability definition name 1> :
>     type: tosca.capabilities.Container
>     valid_source_types: [tosca.nodes.SoftwareComponent]
>   <capability definition name 2>:
>     type: tosca.capabilities.Endpoint.Admin
>
>
> 1)  I am assuming capability definition name 1 and capability definition
> name 2 must be different. Do you agree?
>
> 2) I am assuming requirement definition name 1 and capability definition
> name 1 can be the same. Do you agree?
>
> -Steve
>
> -----Original Message-----
> From: Tal Liron [mailto:[email protected]]
> Sent: Thursday, October 26, 2017 2:37 PM
> To: [email protected]
> Subject: Re: Definition names
>
> I'm not entirely sure what you mean... here's a reply according to what I
> understand.
>
> In most cases there is no ambiguity because the language is YAML-based. A
> dict in YAML has unique keys, so it follows that definitions would be
> unique. (We discussed the issue of importing in a previous thread, and
> opened a JIRA for it.)
>
> There is one curious exception: sequenced lists. In node templates, you
> define requirements as a sequenced list, meaning that you are allowed to
> specify the same requirement name multiple times. This makes perfect sense,
> and matches the "occurrences" field in the the node type.
>
> However ... at the node type the requirement definition is *also* a
> sequenced list. There is no technical reason for this: I believe TOSCA
> defined it this to make it match the node template style, though in my
> personal opinion this was a wrong choice because by definition the
> requirement is unique per node type. ARIA treats the requirements sequenced
> list in the node type the same way the YAML parser treats keys in a dict:
> if there is a key of the same name, it will overwrite a previous key of
> that name.
>
> On Thu, Oct 26, 2017 at 1:04 PM, Steve Baillargeon <
> [email protected]> wrote:
>
> > Hi
> > The TOSCA YAML Profile is not 100% clear about duplicate definitions
> names.
> > What are the guidelines for ARIA?
> >
> > Should all definitions names be unique across definitions "classes"
> > (attributes, properties, requirements, capabilities,...) within a
> > given node type definition?
> > Or is it OK to only have unique definitions names within a given
> > definition "class"?
> >
> > Regards
> > Steve B
> >
> >
>

Reply via email to