mbeckerle commented on a change in pull request #143: Support 
dfdl:length="prefixed"
URL: https://github.com/apache/incubator-daffodil/pull/143#discussion_r239913945
 
 

 ##########
 File path: 
daffodil-core/src/main/scala/org/apache/daffodil/grammar/AlignedMixin.scala
 ##########
 @@ -125,8 +126,8 @@ trait AlignedMixin extends GrammarMixin { self: Term =>
   }
 
   private lazy val priorAlignmentApprox: AlignmentMultipleOf = {
-    if (this.isInstanceOf[Root]) {
-      AlignmentMultipleOf(0) // root is aligned with anything
+    if (this.isInstanceOf[Root] || this.isInstanceOf[DetachedElementDecl]) {
+      AlignmentMultipleOf(0) // root and detached elements are aligned with 
anything
 
 Review comment:
   I think some of this stuff is wrong in the spec. But let's implement what it 
says there, and then consider correcting it.
   
   I believe none of the properties on an element having lengthKind 'prefixed' 
should affect the representation of the prefix itself.  Properties like 
prefixIncludesPrefixLength affect the interpretation of the value from that 
prefix, but not the representation of the prefix. 
   
   I believe all  properties that are supposed to apply to the prefix itself 
must be expressed on a simple type definition. There should be no property 
combining between the prefix's simple type, i.e., the pseudo-element you create 
for it, and the element having that prefix type. 
   
   Let's take the case of a string with a 4-byte binary integer as prefix 
length. The integer has to be 32-bit aligned.
   
   Properties on the element want to be those associated with a textual string, 
not those for a binary integer. 
   
   Consider a harder case: Consider the prefix length is itself a text integer, 
giving the length of another text integer. Then if the property 
dfdl:textNumberJustification and dfdl:textTrimKind, etc. appear on the element, 
do those apply to the element, the prefix or both. I claim they apply ONLY to 
the element, NOT to the prefix. Any properties for the prefix have to be 
expressed on the simple type definition, which will have to be a named type 
definition someplace if any properties are needed.
   
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

Reply via email to