[
https://issues.apache.org/jira/browse/JCR-1239?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12546197
]
Marcel Reutegger commented on JCR-1239:
---------------------------------------
I think we should keep the SPI interface as simple as possible and not assume
too much about the ability of an implementation. The size of the child node
entries collection might be expensive to calculate for an implementation.
Even with the current Iterator the size can be easily counted in jcr2spi for a
reasonable amount of child nodes (i.e. 1000). This is the most common use case
and will provide a correct RangeIterator.getSize(). For really large
collections (e.g. > 10'000) of child nodes I suggest the getSize() in jcr2spi
should return -1. Even though not currently implemented, the jcr2spi layer
should not cache these anyway because of memory consumption.
Therefore I propose to close this as won't fix.
> SPI: Return value of RepositoryService.getChildInfos
> ----------------------------------------------------
>
> Key: JCR-1239
> URL: https://issues.apache.org/jira/browse/JCR-1239
> Project: Jackrabbit
> Issue Type: Improvement
> Components: jackrabbit-spi
> Reporter: angela
> Fix For: 1.4
>
>
> separate issue started by julian within issue #1166:
> RepositoryService.getChildInfos is defined to return an Iterator.
> this forces the caller to loop over all entries in order to detect the number
> of child-nodes.
> currently the hierarchy is populated any way, but that could be a source for
> future optimization.
> possible solutions:
> - change return value to RangeIterator
> - change return value to Collection (or Set).
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
