| This is totally optional and experimental. The only rational is that
| for large documents or documents with long tag names, this saves a lot
| of bytes.
Tests have shown that this difference disappears under compression.
| Think of it as a kind of compression technique that would
| only be enabled when both ends of the pipe can handle it.
Empty end tags are a well formedness error, and the behavior of a
conforming XML processor upon encountering such an error is to stop
parsing.
The prohibition on empty end tags was adopted specifically to enable
users to perform a large class of maintenance operations on XML
documents without having to buy commercial software. I'm having a
very difficult time seeing this as anything but a blatant attempt to
subvert the standard by implementing a nonstandard feature in a widely
disseminated parser. Please help me to understand this differently.
Jon