[ 
https://issues.apache.org/jira/browse/DAFFODIL-1909?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16374428#comment-16374428
 ] 

Steve Lawrence commented on DAFFODIL-1909:
------------------------------------------

The way the InfosetInputter currently works is it maintains state for the 
current infoset event (i.e. start/end element/document). The Daffodil 
implementation calls functions to request the pieces of state (e.g. 
getLocalName, getNamespaceuRI, getSimpleText) for that infoset event and build 
the internal Inofset nodes that Daffodil needs. Once Daffodil has gotten all 
the information it needs, it calls next() to to tell the InfosetInputter to 
move to the next infoset event.

So it's more of a "pull" API than a "push" API. I'm not against changing this, 
but as is, I think the trait can still be readOnly.

> JAPI needs DISimple, DIComplex, DIArray, DIElement, etc for InfosetOutputter
> ----------------------------------------------------------------------------
>
>                 Key: DAFFODIL-1909
>                 URL: https://issues.apache.org/jira/browse/DAFFODIL-1909
>             Project: Daffodil
>          Issue Type: Bug
>          Components: API
>    Affects Versions: 2.1.0
>            Reporter: Michael Beckerle
>            Priority: Major
>             Fix For: 2.2.0
>
>
> Our JAPI and SAPI don't allow one to drill down (into the javadoc/scaladoc) 
> of the DISimple and other DINode subclasses that are passed to the 
> InfosetOutputter methods.
> The information on these DINode classes, including the ElementRuntimeData 
> information needed by an infoset outputter needs to be on documented APIs.
> This suggests we need to revisit the DINode classes and subclasses, and 
> create a read-only trait that we cast up to when passing these to the 
> InfosetOutputter API methods. 
> This is what InfosetElement, InfosetDocument, and InfosetArray were supposed 
> to be, though whether they are suitable is unclear. The DISimple/DIComplex 
> distinction, for example, is very helpful when creating an InfosetOutputter. 
> We may also want to provide an interface to a subset of the ERD data - as 
> many of its members are internally about the needs of parsers/unparsers and 
> not relevant to the external API for Daffodil.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to