> From: JP Bader [mailto:j...@zavteq.com] > Sent: 23 February 2012 17:10 > > The only question/reference point I have is the lack of requirement for this in several other > languages I've recently been learning, i.e. Ruby, Python, Scala (IIRC). They can compile > just fine with or without ';'. Is it really that important to make this required? Scala has the same sloppy approach to semicolons as AS3 and so suffers from similar problems when the compiler does and doesn't infer semicolons wrongly, eg see http://stackoverflow.com/questions/2246212/why-does-scalas-semicolon-inferen ce-fail-here. In this case "well Scala does it badly too" seems a poor argument.
Ruby (as far as I know) does things properly. A newline always denotes the end of an expression. So BNF can precisely depict the language as ";" and "\n" are equivalent. Semicolons aren't really optional in Ruby, it just makes no sense to put them at the end of a line. I don't know Python, so cannot comment on that. When a language uses sloppy optional semicolons, one is forced down one of two paths with a parser: hand craft a parser (the approach used for mxmlc I believe) or hack a BNF-based one to handle the semicolons (the approach taken by Falcon.) I'd prefer to avoid both of these methods and instead to come up with a precise BNF definition of AS3 (or as near to AS3 as BNF can get.) Doing so will make writing the parser far easier and gives us a parser that can be expressed in a meta language. A whole range of different language-based compilers (C, C#, Java, etc) can be generated from the one source. David. JP On Thu, Feb 23, 2012 at 6:53 AM, saurabh jain <jainsaurab...@gmail.com> wrote: > +1 to mandatory semicolon. > I always thought why this was not mandatory. > > Regards, > Saurabh -- JP Bader Principal Zavteq, Inc. @lordB8r | j...@zavteq.com 608.692.2468
