[wg-camlp4] Against the use of syntactically-valid OCaml code for syntax extension purposes

[Alain Frisch]

On 01/30/2013 09:36 AM, Leo White wrote:
> An extension could only support {: foo } style attributes (the ones that
> the type-checker fails on), so then there would be no risk of a typo or
> missed ppx option causing a silent fail or an unhelpful error message.

The risk is that the user writes {@ foo} instead of {@ foo}.  This does 
not sound much less probable than the user writing {@ foox} instead of 
{: foo}.  But I will not fight against non-ignorable attributes!

> For instance, "foo bar {: ... }" means that foo will be called with bar
> and whatever that attribute becomes as arguments.

This was not my understanding.  You seem to imply that {: ... } here is 
not really a (postfix) attribute on the expression "foo bar", but that 
it is a stand-alone expression.  So actually, it is not attached on 
anything, it is not an attribute. I'd rather write it:

   foo bar ((@marker) e)

where the @marker attribute is to be detected by the "expander" and e is 
its argument.  If we want to make it clear that "e" will be interpreted 
in a non-trivial way and that (@marker) must be treated (and otherwise 
the type-checker should complain by mentioning it explicitly), we could 
write:

  foo bar ((:marker) e)

(Assuming now that (:...) is in prefix position.)


Alain