Comment Syntax

[Manuel M T Chakravarty]

Josef Svenningsson:
> I'm in favour of changing the comment syntax.
> 
> On 2/2/06, Manuel M T Chakravarty <chak at cse.unsw.edu.au> wrote:
>         I am against such a change.  The change would break existing
>         software
>         (eg, Yampa) and secondly I don't buy the "main sources of 
>         confusion for beginners" argument.  The confusion arises only
>         when a
>         single line comment is used to uncomment a set of characters
>         that start
>         with a special symbol.  That's a situation that doesn't arise
>         that 
>         often.  (I'd actually be very happy if the main sources of
>         confusion fpr
>         beginners where of such simple syntactic nature.)
> 
> Oh yes, it does happen that a single line comment begins with a
> special symbol. It has happened to me on several occations when using
> haddock annotation to my source code. It is all to easy to forget that
> extra space. With incomprehensible error messages as a result. 

I didn't say it doesn't happen.  I said, it doesn't happen that often.
Haddock increases the likelihood of it happening, but as Henrik wrote,
well just improve the errors messages a bit.  Errors involving operators
starting with -- could specifically suggest that the user might have
wanted a comment, but wrote an operator.

>         As for consistency, well if you absolutely want to make it
>         consistent,
>         impose the same rule on {- as on --.
> 
> I still think there is an inconsistency here. And it has to do with
> maximal munch lexing. Maximal munch is what we normally expect from a
> lexer for a programming language. But the way comments work at the
> moment breaks maximal munch. The longest possible read is to read the
> whole line as a comment and not interpret for instance --^ as an
> operator. It breaks any programmers' intuition not only beginners'. I
> still get it wrong from time to time. 

That doesn't convince me either.  Comment syntax breaks standard lexical
analysis for nested comments anyway (as they are not regular).

And as I wrote before, if we were to design a language from scratch, I
might be persuaded to change my opinion, but if we change the rule for
Haskell now, we break good code.  I believe, we should not break good
code without a really good reason.  I haven't seen a really good reason
to change the status quo.

Manuel