[
https://issues.apache.org/jira/browse/ARIA-149?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16021272#comment-16021272
]
ASF GitHub Bot commented on ARIA-149:
-------------------------------------
Github user ran-z commented on a diff in the pull request:
https://github.com/apache/incubator-ariatosca/pull/138#discussion_r118011107
--- Diff: aria/modeling/service_instance.py ---
@@ -1728,34 +1732,41 @@ def operation_template(cls):
def inputs(cls):
return relationship.many_to_many(cls, 'parameter',
prefix='inputs', dict_key='name')
+ @declared_attr
+ def configuration(cls):
+ return relationship.many_to_many(cls, 'parameter',
prefix='configuration', dict_key='name')
+
+ @declared_attr
+ def arguments(cls):
+ return relationship.many_to_many(cls, 'parameter',
prefix='arguments', dict_key='name')
+
# endregion
description = Column(Text)
relationship_edge = Column(Boolean)
implementation = Column(Text)
- configuration = Column(modeling_types.StrictDict(key_cls=basestring))
dependencies = Column(modeling_types.StrictList(item_cls=basestring))
+ function = Column(Text)
executor = Column(Text)
max_attempts = Column(Integer)
retry_interval = Column(Integer)
def configure(self):
- from . import models
- # Note: for workflows (operations attached directly to the
service) "interface" will be None
- if (self.implementation is None) or (self.interface is None):
+ if (self.implementation is None) and (self.function is None):
return
- if self.plugin is None:
- arguments =
execution_plugin.instantiation.configure_operation(self)
+ if (self.plugin is None) and (self.interface is not None):
+ # Default to execution plugin ("interface" is None for
workflow operations)
+ execution_plugin.instantiation.configure_operation(self)
else:
# In the future plugins may be able to add their own
"configure_operation" hook that
- # can validate the configuration and otherwise return
specially derived arguments
- arguments = self.configuration
+ # can validate the configuration and otherwise create
specially derived arguments. For
+ # now, we just send all configuration parameters as arguments
+ utils.instantiate_dict(self, self.arguments,
self.configuration)
- # Note: the arguments will *override* operation inputs of the same
name
- if arguments:
- for k, v in arguments.iteritems():
- self.inputs[k] = models.Parameter.wrap(k, v)
--- End diff --
why remove this?
> Support instrinsic functions in "dependencies" operation configuration
> ----------------------------------------------------------------------
>
> Key: ARIA-149
> URL: https://issues.apache.org/jira/browse/ARIA-149
> Project: AriaTosca
> Issue Type: Story
> Reporter: Tal Liron
> Assignee: Tal Liron
>
> There are a few issues here. First, the following YAML is totally broken;
> {code}
> dependencies:
> - ssh.address > { get_attribute: virtual_ip.floating_ip }
> {code}
> The problem is that in YAML, due to the location of the ":", this gets parsed
> as a dict where the key is {code}ssh.address > { get_attribute{code} and the
> value is {code}virtual_ip.floating_ip }{code} which is not what we want at
> all.
> The solution is to encase the whole value in quotes to enforce parsing as a
> string:
> {code}
> dependencies:
> - "ssh.address > { get_attribute: virtual_ip.floating_ip }"
> {code}
> Note that the one attractive solution is to to change this ">" format to a
> real dict. So:
> {code}
> dependencies:
> - { ssh.address: { get_attribute: virtual_ip.floating_ip } }
> {code}
> But this will break other TOSCA parsers that expect a string here, so no go.
> :(
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
