[
https://issues.apache.org/jira/browse/OAK-4994?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16161104#comment-16161104
]
Michael Dürig commented on OAK-4994:
------------------------------------
Linking to OAK-6584 as the additional record types are likely most useful for
record level tooling.
> Implement additional record types
> ---------------------------------
>
> Key: OAK-4994
> URL: https://issues.apache.org/jira/browse/OAK-4994
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: segment-tar
> Reporter: Francesco Mari
> Labels: tooling
>
> The records written in the segment store should be augmented with additional
> types. In OAK-2498 the following additional types were identified:
> - List of property names. A list of strings, where every string is a property
> name, is referenced by the template record.
> - List of list of values. This list is pointed to by the node record and
> contains the values for single\- and multi\- value properties of that node.
> The double indirection is needed to support multi-value properties.
> - Map from string to node. This map is referenced by the template and
> represents the child relationship between nodes.
> - Super root. This is a marker type identifying top-level records for the
> repository super-roots.
> Just adding these types doesn't improve the situation for the segment store,
> though. Bucket and block records are not easily parseable because they have a
> variable length and their size is not specified in the record value itself.
> For record types to be used effectively, the way we serialize certain kind of
> data has to be reviewed for further improvements.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
