[jira] [Commented] (DAFFODIL-2824) warning needed when all contents are optional inside an optional element and dfdl:emptyElementParsePolicy="treatAsEmpty"

["Mike Beckerle (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DAFFODIL-2824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17732705#comment-17732705 ] 

Mike Beckerle commented on DAFFODIL-2824:
-----------------------------------------

[~olabusayo] please proofread the above description. I want to be sure this captures the issue we were looking at today. 

> warning needed when all contents are optional inside an optional element and dfdl:emptyElementParsePolicy="treatAsEmpty"
> ------------------------------------------------------------------------------------------------------------------------
>
>                 Key: DAFFODIL-2824
>                 URL: https://issues.apache.org/jira/browse/DAFFODIL-2824
>             Project: Daffodil
>          Issue Type: Improvement
>          Components: Middle "End"
>    Affects Versions: 3.4.0
>            Reporter: Mike Beckerle
>            Priority: Major
>
> A common mistake is to have an optional element of complex type which is a sequence of all optional elements.
> This requires dfdl:emptyElementParsePolicy="treatAsAbsent", which is NOT the usual default in Daffodil. 
> (It is the default in the IBM-portable formats, just not in the standard daffodil-only formats)
> Example: (assume context is delimited)
>  
> {code:java}
> <sequence dfdl:separator="%WSP+">
>     <element name="required1" type="w:zlString" dfdl:initiator="999"/>
>     <element name="optParent" minOccurs="0">
>          <complexType>
>             <sequence>
>               <element name="optChild1" minOccurs="0" dfdl:initiator="//" type="w:string3"/>
>               <element name="optChild2" minOccurs="0" dfdl:initiator="[[" type="w:string3" dfdl:terminator="]]"/>
>            </sequence>
>         </complexType>
>      </element>
>      <element name="optional2" type="zlString" dfdl:initiator="opt2" minOccurs="0"/>
> </sequence>
> {code}
> In the example above, if the data is
>  
> {noformat}
> 999
> opt2
> {noformat}
> We want the infoset to be 
>  
> {code:java}
> <required1/>
> <optional2/>
> {code}
> But it won't be. It will try to construct this infoset:
>  
> {code:java}
> <required1/>
> <optParent/>
> {code}
>  
> But you can see that the optParent is deemed present, because an empty element is perfectly allowed by this schema, and dfdl:emptyElementParsePolicy="treatAsEmpty".
> Even though none of its children are present. It will then expect another separator, and will fail because there isn't one. 
> To fix this, one should set dfdl:emptyElementParsePolicy="treatAsAbsent", and that's what the warning message should suggest. 
> A less attractive fix is to add an assertion at the end of the optParent element like this:
>  
> {code:java}
>     <element name="optParent" minOccurs="0">
>          <complexType>
>             <sequence>
>             <element name="optChild1" minOccurs="0" dfdl:initiator="//" type="w:string3"/>
>             <element name="optChild2" minOccurs="0" dfdl:initiator="[[" type="w:string3" dfdl:terminator="]]"/>
>              <sequence>
>                 <annotation><appinfo soruce="http://www.ogf.org/dfdl/"/>
>                     <!-- 
>                       One of the above children must exist or the
>                       enclosing parent does not exist
>                       -->
>                     <dfdl:assert>{ fn:exists(optChild1) or fn:exists(optChild2) }</dfdl:assert>
>                </appinfo></annotation>
>              </sequence>
>            </sequence>
>         </complexType>
>      </element>
> {code}
>  
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)