Re: Synopsis as documentation tool

Gilles J. Seguin wrote:
> On Tue, 2005-12-13 at 00:42 -0800, TR wrote:
> 
>>Seems pretty ambitious. I'd be interested in hearing a description of how such 
>>a thing might be accomplished.


> Synopsis is not ready for this yet.
> - missing label list in function-body production.
> - scopes introduce by declaration in block statement.  You can not jump
>   in a middle of a scope.

There are a number of analysis steps missing, but they can be worked on
one step at a time:

* type analysis

* overload resolution
   (at this point you will be able to construct a call graph)

* constant expression evaluation
   (at this point explicit template instantiation is
   possible, allowing the second half of two-phase symbol lookup)

* ...

I think nobody doubts that these are complex tasks, but I believe there are great
benefits to have support for these features in a tool like synopsis.

Regards,
		Stefan