This is a reminder that we're having a DFDL Workgroup call this Thursday June 11, 11am
EDT.US.
Please send items you'd like on the agenda, and let me know if you plan to attend.
I have this topic for discussion:
An idea has been proposed to generalize this to a more flexible peek/look-ahead capability.
This is a requested feature for DFDL 2.0 allowing a parser to learn from examining the data once, and then use that knowledge in a subsequent parse over the same data.
Details:
- annotation on a xs:sequence
- a parser to speculate forward and when it reaches the end of the sequence, resetting the position to the start of the sequence, but retaining information from the parse.
- the retained information certainly includes DFDL variables set during the peek-ahead parse.
- more controversial is the ability to retain the infoset, effectively allowing one to parse the same data multiple times.
- does nothing on unparse. The annotated sequence is simply skipped by the unparser.
- seems very easy to implement.
Mike Beckerle