We have ways of cheating a bit if we want to reinterpret the semantics of something that nevertheless parses cleanly - while the parser is limited to single token lookahead, it's straightforward for the subsequent code generation stage to look a single level down in the parse tree and see that the code that parsed as "with expr" is actually "with subexpr as target".

It does work, however I think it does sound like a patch, and definitely it will block us to make other extensions in the future.

3) "target ? expr" (where ? is some other word/character - IIRC "target from expr" was proposed once)

A more popular convention is to mark `?` as handling boolean variables, so `target ? expr` could mean `expr if target else target`. Other proposal for null/boolean checking might need `?`, let's preserve `?` character for further development.

How about "name being expression" - this avoids the already used "as" while being searchable, reasonably short and gives a reasonably clear, (at least to English speakers), indication of what is going on. It can also be typed on an ASCII keyboard without having to have a helper program or memorising Unicode codes and can be displayed or printed without having to install specialised fonts.

It makes sense, if we don't have a long history in Python programming... A new keyword would be something very dangerous, because it just causes the crash of some existed library using the keyword as identifier.