[
https://issues.apache.org/jira/browse/DATAFU-50?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Eyal Allweil updated DATAFU-50:
-------------------------------
Attachment: DATAFU-50-2.patch
Hi guys - [~mrflip], [~matterhayes] - what I suggest is merging this without
the change in implementation for AliasableEvalFunc. That will add the
capabilities that were proposed in this jira's description without needing to
modify all the Bag UDF's to preserve their existing behavior, which is what led
to the test failures.
This will also allow us to merge
[DATAFU-47|https://issues.apache.org/jira/browse/DATAFU-47].
I think I've managed to tweak the patch properly to do this - all tests pass
now.
> SimpleEvalFunc should extend ContextualEvalFunc, have good lifecycle hooks
> --------------------------------------------------------------------------
>
> Key: DATAFU-50
> URL: https://issues.apache.org/jira/browse/DATAFU-50
> Project: DataFu
> Issue Type: Improvement
> Reporter: Philip (flip) Kromer
> Assignee: Philip (flip) Kromer
> Labels: Utils
> Attachments:
> 0001-DATAFU-50-SimpleEvalFunc-should-extend-ContextualEva.patch,
> DATAFU-50-2.patch
>
>
> Being able to squirrel away values frozen at frontend-launch-time is very
> handy.
> This does two things:
> * makes SimpleEvalFunc extend ContextualEvalFunc rather than
> plain-old-EvalFunc. This seems harmless and gains you all that power of
> Contextualism.
> * Adds a hook method "void schemaReadyHook(inputSchema,outputSchema)" to
> ContextualEvalFunc, called once the schemata are prepared. This acts as sugar
> for classes that want to set properties at front-end time to be used at later
> points but which, like most SimpleEvalFunc's, want nothing to do with schema
> calculation. In ContextualEvalFunc it does nothing; subclasses may override
> it and if so must call super. schemaReadyHook().
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
