Introducing Cobra

[SJS]

begin  quoting Christopher Smith as of Wed, Jan 02, 2008 at 02:07:17PM -0800:
> John H. Robinson, IV wrote:
[snip]
> > I agree Python's approach is wrong. Whitespace should not be
> > syntactically significant, outside of separating tokens.

Whitespace is a boolean: "Is there whitespace here?" 

> > I understand the only two valid arguments:
> > 1) If you remove the braces from the different bracing style, you are
> >    left with Python-mandated indenting.
> > 2) There is no question which IF block the ELSE is associated with.
>
> There is a third valid argument: the eye of the reader tends to
> "believe" whitespace is significant; more significant than braces (which
> is why people use whitespace in the first place). So, rather than fight
> that, let it mean what the eye thinks it means and you improve readability.

Except that it *doesn't* improve readability.

The eye *believes* whitespace, but it's also not very good at it.  It's
okay at determining that yeah, verily, there *is* whitespace, but it's
not very good at determining how much.  For very small programs, that's
okay; but these languages are no longer being used solely for small programs.

Now, if only three levels of indentation were allowed (none, some, a lot),
then sure, the eye would be really good at figuring that out. But is that
really the sort of indentation constraint we want to see in a language?

> The problem comes from cases where people use a mix of tabs and spaces,
> particularly if there are non-standard tab sizes to deal with. Then what
> the eye sees and what the compiler see don't match again.

There are several problems. That's just one of 'em.

-- 
I could contrive a basic example apace
But I'm afraid it won't fit this space
Stewart Stremler