[bitc-dev] On whitespace sensitivity (was Re: Opinions wanted: Infix shift operators)

[wren ng thornton]

William Leslie wrote:
>> When programming in Python, my intuitions are constantly thwarted in one
>> way or another. E.g., requiring line-final backslashes to do line
>> wrapping is a gross hack;
> 
> If you need to use that you probably have other problems. Line
> continuation is implicit inside parenthesis, braces and brackets,
> which is the only place you'd ever want them.

Not in my experience. Why should I add extraneous parentheses around a 
top-level expression, like a summation or product of a bunch of 
parameters, as in statistical number crunching? For casual scripting 
where most of your statements are small atomic actions, Python suffices; 
but whenever the program becomes mathematically intricate or requires 
descriptive names, it's very easy to hit the 80 character limit without 
involving parens, braces, or brackets.


> But then you are really talking about python treating newline as a
> statement delimiter, not about indentation. I think that newline makes
> a very intuitive statement delimiter.

It's all the same thing: the use of whitespace as semantically 
meaningful for determining the boundaries of statements and blocks. As I 
said earlier, I think Haskell's solution is far more successful and 
intuitive than simply replacing the semicolon or period with a \n.

-- 
Live well,
~wren