You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@avro.apache.org by "Todd Lipcon (JIRA)" <ji...@apache.org> on 2010/02/05 23:40:28 UTC
[jira] Created: (AVRO-404) Support "mixin properties" for specific
codegen
Support "mixin properties" for specific codegen
-----------------------------------------------
Key: AVRO-404
URL: https://issues.apache.org/jira/browse/AVRO-404
Project: Avro
Issue Type: New Feature
Components: java, spec
Reporter: Todd Lipcon
As we add more features for RPC and the specific codegen, there are some codegen options that should be provided as annotations on schema fields. For example, some users may want to turn binary fields into byte[], and other people may prefer ByteBuffer. Since these are codegen-scoped and not data-scoped, it would be nice to allow the properties to be "mixed in" at codegen time. By moving them out of the main schema definition, different users of the same Avro protocol could make different choices for the codegen in their specific applications.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (AVRO-404) Support "mixin properties" for
specific codegen
Posted by "Philip Zeyliger (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/AVRO-404?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12830467#action_12830467 ]
Philip Zeyliger commented on AVRO-404:
--------------------------------------
I think I know what Todd is talking about here, so I'll do an example.
We might have a record X. One user might want to annotate it with "java-class: org.foo.X". Another user might want to also use the reflection API, but annotate it with "org.bar.X". Since X has been defined externally (in some other project, say), nobody really wants to fork X, both 'foo' and 'bar' users merely want to add some mix-ins. They could certainly do so at runtime, but then the schema is this weird "mixed-mode" thing. That's the reflect API example. The specific API example happens when we end up having options about how we generate the generated code, and different users want different options.
I think we might be able to make this more of a "genavro"/(avroidl?) issue rather than a schema issue. Support some sort of modification of existing schemas after inclusion...
> Support "mixin properties" for specific codegen
> -----------------------------------------------
>
> Key: AVRO-404
> URL: https://issues.apache.org/jira/browse/AVRO-404
> Project: Avro
> Issue Type: New Feature
> Components: java, spec
> Reporter: Todd Lipcon
>
> As we add more features for RPC and the specific codegen, there are some codegen options that should be provided as annotations on schema fields. For example, some users may want to turn binary fields into byte[], and other people may prefer ByteBuffer. Since these are codegen-scoped and not data-scoped, it would be nice to allow the properties to be "mixed in" at codegen time. By moving them out of the main schema definition, different users of the same Avro protocol could make different choices for the codegen in their specific applications.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (AVRO-404) Support "mixin properties" for
specific codegen
Posted by "Doug Cutting (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/AVRO-404?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12830347#action_12830347 ]
Doug Cutting commented on AVRO-404:
-----------------------------------
The reflect API uses the "java-class" property in a schema for a related purpose. It also uses annotations to specify this.
> Support "mixin properties" for specific codegen
> -----------------------------------------------
>
> Key: AVRO-404
> URL: https://issues.apache.org/jira/browse/AVRO-404
> Project: Avro
> Issue Type: New Feature
> Components: java, spec
> Reporter: Todd Lipcon
>
> As we add more features for RPC and the specific codegen, there are some codegen options that should be provided as annotations on schema fields. For example, some users may want to turn binary fields into byte[], and other people may prefer ByteBuffer. Since these are codegen-scoped and not data-scoped, it would be nice to allow the properties to be "mixed in" at codegen time. By moving them out of the main schema definition, different users of the same Avro protocol could make different choices for the codegen in their specific applications.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.