I added commits to the PR on the base64/layering feature. https://github.com/apache/incubator-daffodil/pull/61
Unparsing that is symmetric with parsing now works. There's only a few tests, but they do include one that gzips text, measuring how big it turns out to be, stores that length followed by the gzipped content, then base64's that pair of things. I think it is worth re-review at this point. My next step is adding line-folding and some richer examples that do things more like what is found in IETF data formats. I will be doing this with TDML tests in the daffodil-test directory, as opposed to the unit tests I've done thusfar.
