> In the Haskell rules as I understand them, semicolon auto-insertion is tied 
> to curly brace insertion. If a programmer hand-inserts the curly brace, they 
> are necessarily taking responsibility for the semicolons.

Yes.

> But automatic semicolon insertion could be done even when the curly brace has 
> been inserted by hand. It simplifies the layout state machine a bit if this 
> is done. Is there a reason not to do so?

It sounds like you're thinking about this from an implementor's
perspective.  But think about it from a programmer's perspective.
Why would you want to have semicolons inserted automatically in
code that has explicit curly braces around it?  An opening brace
is a programmer's way of saying "I don't want to be constrained
in how I lay out this code", "I don't want to have to think about
layout", "I object to the idea that whitespace has significance",
or "This code was generated by a program that doesn't know about
Haskell's layout rule".  Auto-inserting semicolons in cases like
these would defeat the purpose of having explicit braces.

Personally, I'd be happy if implicit layout was the only option
(no explicit braces)---I never have any of the concerns mentioned
above.  This would also simplify your state machine :-)  However,
not everyone in the Haskell community shares my enthusiasm for
layout.

All the best,
Mark


_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to