2.0.0 Release Candidate 1
2.0.0 Release Candidate 1
Optimizing out the capture parser/unparsers for elements that have no
expressions referencing them via valueLength or contentLength

This means that for pcap parsing we should ONLY be calling the
capture parsers for content length or value length where absolutely
necessary.

For unparsing, note that there are a lot of reasons why we capture the value
region length however. E.g., alignment, mandatory text alignment are the two
real biggies.

Until we optimize out alignment and mandatory text alignment there are
still going to be a fair number of these capture processors put in by
the compiler.

Remove slow downcast on critical path.

Added pooling for DIElementState objects.

On demand creation and copying of ContentLengthState and
ValueLengthState objects.

Removed return statements from all repeating parse combinators. Hoping
this might improve performance a little. Removed try/finally from rep
parsers and alt-comp parser.

Removed println for debug left in on purpose in earlier commits.

DFDL-1610, DFDL-1613
53 files changed