Django is very forgiving on incomplete tags, reverting to treat them as just ordinary text.
Related to #103.
I think the scanner/parser interaction needs a major refactoring in order to support this, so it's more of a back-and-forth as you go kind of thing rather than a scan-all-then-parse-it.
The upside is that it would be a lot easier to dynamically add tags in such a scenario.
Let's see if it is at all feasible..