[
https://issues.apache.org/jira/browse/DAFFODIL-2410?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17213999#comment-17213999
]
Mike Beckerle commented on DAFFODIL-2410:
-----------------------------------------
At minimum we need the ability for tests, or whole test suites, to assert that
there should be no warnings. Our current warnings feature lets a test insist on
specific warning text, but we have no mechanism for precluding them.
I would suggest tdml:parser/unparserTestCase gets a allowWarnings="yes/no"
attribute, which combines with the <tdml:warnings> element to allow/preclude
additional warnings beyond those explicitly required. tdml:testSuite should get
defaultAllowWarnings to set this for the whole suite.
For now the default for defaultAllowWarnings would have to be "yes" for
compatibility, but eventually we want to flip this.
We may want a TDML-Runner specific "warning" telling you your tdml:testSuite
does not specify this defaultAllowWarnings and that the preferred setting for
it is "no".
> TDML Runner hides warnings
> --------------------------
>
> Key: DAFFODIL-2410
> URL: https://issues.apache.org/jira/browse/DAFFODIL-2410
> Project: Daffodil
> Issue Type: Improvement
> Components: TDML Runner
> Reporter: Steve Lawrence
> Priority: Major
>
> The TDML Runner currently hides warnings. This means that if we make a change
> that causes many new warnings it is very easy to not realize it.
> Unfortunately, many tests, including those external to daffodil, might rely
> on this current behavior so we can't simply fail tests if warnings exist but
> there are no tdml:warnings in the TDML test.
> However, we should at least have some mode that allows us to run the TDML
> runner and error if there are unchecked warnings. Or perhaps we always output
> warnings, which would encourage users to update their schemas or fix tests to
> suppress expected warnings. We could even provide a message that says in some
> future release uncheck warnings will always be considered an error if we want
> to go that route.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
