[
https://issues.apache.org/jira/browse/ARIA-42?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15858116#comment-15858116
]
ASF GitHub Bot commented on ARIA-42:
------------------------------------
Github user mxmrlv commented on a diff in the pull request:
https://github.com/apache/incubator-ariatosca/pull/68#discussion_r100087789
--- Diff: aria/storage/sql_mapi.py ---
@@ -364,6 +369,28 @@ def _load_relationships(instance):
getattr(instance, rel.key)
+def init_storage(base_dir=None, filename='db.sqlite'):
+ if base_dir is not None:
+ uri = 'sqlite:///{platform_char}{path}'.format(
+ # Handles the windows behavior where there is not root, but
drivers.
+ # Thus behaving as relative path.
+ platform_char='' if 'Windows' in platform.system() else '/',
+
+ path=os.path.join(base_dir, filename))
+ engine_kwargs = {}
+ else:
+ uri = 'sqlite:///:memory:'
--- End diff --
remove in memory option to tests only
> Implement full blown mechanism for serializing an operation context
> -------------------------------------------------------------------
>
> Key: ARIA-42
> URL: https://issues.apache.org/jira/browse/ARIA-42
> Project: AriaTosca
> Issue Type: Story
> Reporter: Dan Kilman
> Assignee: Maxim Orlov
>
> The current mechanism at
> aria/orchestration/workflows/context/serialization.py is very limited.
> It serializes/deserializes the context.model and context.resource with hard
> coded knowledge of the actual implementation involved and even how they were
> initialized. Specifically:
> The model storage initialized is always expected to be sqlachemy based with
> no additional engine configuration (i.e. only the engine url is used to
> re-construct the sqlalchemy engine), similar logic applies to the
> instantiated sqlalchemy session.
> The resource storage initialized is always the file based resource storage.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
