[
https://issues.apache.org/jira/browse/ACCUMULO-3741?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14500995#comment-14500995
]
Billie Rinaldi commented on ACCUMULO-3741:
------------------------------------------
I do not expect to switch to version 3.2.0-incubating for Accumulo 1.7.0.
3.2.0-incubating has not actually been released yet, but it's being discussed.
However, I'd like to evaluate the changes that would be necessary and see if it
would make sense to bring any of them into 1.7.0. For example, I think the
thrift object RemoteSpan will need to be changed, and in some ways it would be
changing back to how it was in 1.6.0. It would be nice to avoid switching back
and forth. I'll attach the patch I have so far, as a preview. It compiles
with both versions of htrace, but I am only just starting to test it now.
> Reduce incompatibilities with htrace 3.2.0-incubating
> -----------------------------------------------------
>
> Key: ACCUMULO-3741
> URL: https://issues.apache.org/jira/browse/ACCUMULO-3741
> Project: Accumulo
> Issue Type: Bug
> Reporter: Billie Rinaldi
> Assignee: Billie Rinaldi
>
> HTrace 3.2.0-incubating is coming out soon and there are some significant
> differences in its API. Ideally we would like Accumulo to work with either
> version. Differences include:
> * no Trace.setProcessId method
> * process ID is handled differently now; it defaults to name/IP, while before
> Accumulo tracked IP manually
> * SpanReceivers are expected to set process ID if it is not provided
> * return type of Span.getKVAnnotations changed from Map<byte[],byte[]> to
> Map<String,String>
> * Spans can now have multiple parents (although none do yet) and getParentId
> has been replaced by getParents which returns a list of longs
> * multiple parents means there is no longer a fixed root span ID; root spans
> just have an empty list of parents
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
