[
https://issues.apache.org/jira/browse/SOLR-793?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12657765#action_12657765
]
Ryan McKinley commented on SOLR-793:
------------------------------------
| One update request will have only one <add>
That is the existing limitation, but i don't see any reason there could not be
multiple <add> statements within one request. similar to how we have multiple
delete commands in one statement.
Our existing parser supports this already, only we would need to add a new root
element. This would allow a streaming client to post all commands sequentially
to the server.
| The SolrInputDocument can have a 'commitWithin' attribute.
I don't like that because the 'commitWithin' attribute is about the command,
not the data. Attaching it to the 'add' command seems like the logical place
for it.
> set a commit time bounds in the <add> command
> ---------------------------------------------
>
> Key: SOLR-793
> URL: https://issues.apache.org/jira/browse/SOLR-793
> Project: Solr
> Issue Type: Improvement
> Components: update
> Reporter: Ryan McKinley
> Priority: Minor
> Fix For: 1.4
>
> Attachments: SOLR-793-commitWithin.patch,
> SOLR-793-commitWithin.patch, SOLR-793-deadlock.patch
>
>
> Currently there are two options for how to handle commiting documents:
> 1. the client explicitly starts the commit via <commit/>
> 2. set an auto commit value on the server -- clients can assume all documents
> will be commited within that time.
> However, this does not help in the case where the clients know what documents
> need updating quickly and others that could wait. I suggest adding:
> {code:xml}
> <add commitWithin="100">...
> {/code:xml}
> to the update syntax so the client can schedule commits explicitly.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
