Re: [tosca] Proposal for get_property based on discussion in WG on 2019_09_17

[Tal Liron <tliron@redhat.com>]

A long-standing issue for me:

I would also like to propose a special reserved prefix character for function calls. E.g., "$" or "!". So:

{ !get_property: [ ... ] }

The reason should be obvious: as it stands one needs to know all the existing intrinsic functions in order to read a TOSCA source and be sure where there is a function call or just a value of a data type. This will also greatly assist parsers, as they will be able to generate the right error message if it is clear that a function call is intended and the function doesn't exist, etc.

(Puccini allows users to easily create custom functions, something that should probably not be part of the language standard, but that could be acknowledged as a potential feature in the spec.)

On Mon, Sep 30, 2019 at 9:56 PM Chris Lauwers <lauwers@ubicity.com> wrote:

Yes, intrinsic functions deserve some more discussion:

Â

1. get_operation_output has been deprecated
2. get_artifact returns the path to an artifact so it can be passed to some other artifact (presumable a script). There is likely some value in this. However, we also have the concept of a âprimaryâ artifact and âdependentâ artifacts. Itâs not at all clear to me how these are intended to be used.

Â

There are likely other functions that may need to be introduced to make the language more powerful. My understanding is that Heat has a bunch more functions that could be useful in TOSCA.

Â

Chris

Â

From: Tal Liron <tliron@redhat.com>
Sent: Monday, September 23, 2019 5:47 AM
To: Chris Lauwers <lauwers@ubicity.com>
Cc: Calin Curescu <calin.curescu@ericsson.com>; tosca@lists.oasis-open.org
Subject: Re: [tosca] Proposal for get_property based on discussion in WG on 2019_09_17

Â

On Sun, Sep 22, 2019 at 11:09 PM Chris Lauwers <lauwers@ubicity.com> wrote:

-ÂÂÂÂÂÂÂÂÂ I donât believe we can do away with SELF unless we also do away with the ability to specify an arbitrary node in the (containing) topology, as in:

Â

Good point, but then what about getting properties from groups, policies, etc.? get_property always seemed like a poorly designed function to me. For consistency there should be a way to change your starting point to anything.

-ÂÂÂÂÂÂÂÂÂ When discussing your proposal to return sets of values for intrinsic functions, we should start with a discussion about what these functions are intended to be used for.

In a way get_attribute is more interesting, because it involves "live" data.

Â

-ÂÂÂÂÂÂÂÂÂ There are other âintrinsicâ functions in the current spec for which I donât understand the intended us. Specifically, what would we use âget_nodes_of_typeâ for?

Â

What about get_operation_output? And get_artifact? I find a lot of these problematic.