Python 3.5 adds support for the '@' operator, designed for use in matrix multiplication. ("A*B" is element-wise multiplication, and "A@B" is matrix multiplication.)

There's nothing in the implementation which restricts is to matrix multiplication. While there are good reasons to be restrained with operator overloading, I have an idea of a domain-specific use in XML.

XML's XPath defines '@abc' as a way to select an attribute from an element tag. Right now if I want to get all elements named "SEGMENT" and display the "title" and "length" attributes, I might do:

  from xml.etree import ElementTree as ET
  etree = ElementTree.parse("example.xml")
  for segment in tree.find("SEGMENT"):
    print(segment.attrib["title"], segment.attrib["length"])

What about allowing 'x@"y"' as a shorthand for 'x.attrib["y"]', so the last line becomes:

    print(segment@"title", segment@"length")

? Is that too much magic in order to get the domain appropriate syntax?