Entomy
commented
5 years ago I am personally in favor of keeping semicolon termination. The reason being that it's considerably easier on the parser, which isn't just a parser complexity issue, but actually speeds up parsing considerably, while also not requiring much out of the programmer. Of course, statements could simply be line/formatting terminated, but then this is a massive change from the way Ada code is normally written and is considered too extreme of a change.
Consider the case of parsing X := 5 + 4;. We can tell by the second token we are doing an assignment, and everything up until the ; is an expression to evaluate. Because we know we are looking for a ; we can blindly capture everything up until that point, only looking for a ; each position. Once found, not only can the parser continue, but a second parsing thread can evaluate the expression.
Consider the case of parsing X := 5 + 4 where semicolon termination does not happen. Each position we have to check against every possible statement beginning to be sure the end of a statement did not occur. Ada has an absolutely massive amount of statement types, and suddenly the amount of work the parser must do has exploded.
All over one extra keypress.
Ada does: All statements are terminated in semicolons.
This is something contentious and should be discussed, as including them just because of legacy use isn't a very valid argument. However it should be noted the use of semicolons isn't a legacy artifact either.