Ok, lets not argue about coding styles. You still haven't answered fundamental questions such as:
* How exactly does the lexer/parser know when it should insert a soft semicolon?
* How exactly does it give a meaningful error message when it inserts one where the user didn't intend for there to be one?
My problem with your proposal is that it seems to require the parser to apply some complex heuristics to determine when to insert and when to complain meaningfully. It is not obvious to me what these heuristics should be. My suspicion is that such an implementation will just add to perl's already colourful collection of edge cases, and just confuse both beginner and expert alike.
Bear in mind that I am one of just a handful of people who actively work on perl's lexer and parser, so I have a good understanding of how it works, and am painfully aware of its many complexities. (And its quite likely that I would end up being the one implementing this.)
Dave. | 