[
https://issues.apache.org/jira/browse/AVRO-219?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12795869#action_12795869
]
Doug Cutting commented on AVRO-219:
-----------------------------------
> I believe they are broken, but this patch doesn't introduce the problem.
Okay, that's fine then.
> Can we keep the current approach and benchmark it in AVRO-217?
Yes, that's a good plan.
> Your suggestion sounds like a performance optimization to avoid calling
> validate() many times [ .. ]
It is in part performance, but also correctness. A union can contain two
records with different names but which are otherwise identical. The current
definition of validate does not handle this correctly, since it only validates
field names and values and not the record name.
> Rewrite Python implementation's IO path (schema.py, io.py, genericio.py,
> datafile.py) and associated tests
> ----------------------------------------------------------------------------------------------------------
>
> Key: AVRO-219
> URL: https://issues.apache.org/jira/browse/AVRO-219
> Project: Avro
> Issue Type: Improvement
> Components: python
> Reporter: Jeff Hammerbacher
> Assignee: Jeff Hammerbacher
> Attachments: AVRO-219-schema-io-and-datafile.patch,
> AVRO-219.patch.schema, AVRO-219.patch.schema_and_io
>
>
> Currently, the unit tests for schema.py, genericio.py, and datafile.py are
> grouped in with the unit tests for io.py in testio.py. We should break the
> tests into individual files so that we have better modularization of tests.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
