[
https://issues.apache.org/jira/browse/AVRO-1878?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15372153#comment-15372153
]
Ryan Blue commented on AVRO-1878:
---------------------------------
Yeah, I agree that this patch isn't a solution to the problem and it would be
some work to make the resolver responsible for it all. And I do want to
consider whether the alias map should be updated. This caught a couple of
malformed schemas in our own tests, which means that it's a breaking change and
that we should be more strict in the future to make sure behavior is obvious
and well-defined (no guessing which alias wins).
The current behavior is correct, but it's still surprising when passing in two
compatible schemas yields bad data instead of an error. I think that test case
should definitely work, and I think the right design is to keep the resolver
code responsible for all of these. If we all agree in principle on that, then I
can start moving the logic into the resolver with the current patch as starting
place.
> ResolvingGrammarGenerator doesn't handle aliases
> ------------------------------------------------
>
> Key: AVRO-1878
> URL: https://issues.apache.org/jira/browse/AVRO-1878
> Project: Avro
> Issue Type: Bug
> Components: java
> Affects Versions: 1.8.1
> Reporter: Ryan Blue
> Assignee: Ryan Blue
> Fix For: 1.8.2
>
>
> While testing AVRO-1704, I tested schema evolution using an aliased field and
> found that the ResolvingGrammarGenerator doesn't support field aliases.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
