Re: C question (was: Ada & C)

["Ira D. Baxter" <idbaxter@semdesigns.com>]

If you have the right parser technology, parsing C is not hard and you
don't need the symbol table immediately.   GLR parsing handles problems
like typedef names by proposing multiple parses: in this case, one
assuming its a typedef name, and the other assuming its a conventional
identifier.
Which can be resolved later, of course, with the symbol table.

We offer reengineering tools that not only parse C, but C++, Ada95,
Fortran95,
you name it using this technology.   See
http://www.semdesigns.com/Products/DMS/DMSToolkit.html.

I would expect C2ADA to not be very easy.   Pointer casts and arithmetic
are a bitch to translate.


Keith Thompson wrote in message ...
>dvdeug@x8b4e53cd.dhcp.okstate.edu (David Starner) writes:
>> On Wed, 12 Jan 2000 23:38:47 GMT, Ted Dennison <dennison@telepath.com>
wrote:
>> >> [c2ada not working well]
>> >Hmmm. This sounds like a good application for OpenToken. It currently
>> >only ships with Java and Ada syntaxes, but a C syntax shouldn't be too
>> >tough for someone with an ANSI C standard handy to cobble together. I
>> >actually thought C would be the *first* syntax someone submitted, but up
>> >to now no one seems to have been that interested in C. :-)
>> ANSI C's a bit of a bear to parse, or so I've heard.
>
>Indeed.  The biggest problem is typedef names.  When a typedef is in
>scope, the name of the typedef has to be treated as a distinct token,
>not as an identifier -- unless it's a new declaration of the same
>name.  Thus you have to have feedback from the symbol table to the
>parser.  Ick.
>
>--
>Keith Thompson (The_Other_Keith) kst@cts.com  <http://www.ghoti.net/~kst>
>San Diego Supercomputer Center           <*>  <http://www.sdsc.edu/~kst>
>Welcome to the last year of the 20th century.
>