Re: Ada 2012 grammar

[Stephen Leake <stephen_leake@stephe-leake.org>]

David Pereira <dmrpereira@gmail.com> writes:

> Quinta-feira, 24 de Outubro de 2013 17:49:17 UTC+1, G.B. escreveu:
>> On 24.10.13 17:01, David Pereira wrote:
>> 
>> > Does anyone know if there is available a complete grammar for Ada
>> > 2012, lets say, in Flex+Bison or similar compiler construction
>> > tool-chain?
>> 
>> The latest Emacs Ada Mode (5.00) has a Wisent style grammar
>> 
>> for Ada 2012. "Copyright (C) 2012 Free Software Foundation, Inc."
>> 
>> It could be usable after editing it a little.
>
> Dear G.B,
>
> Very nice hint! I am a Emacs user, but didn't occurred that to me
> (shame on me :p). 

It is very new.

Website: http://stephe-leake.org/emacs/ada-mode/emacs-ada-mode.html

The grammar there has conflicts, so it relies on a generalized LALR
parser (http://en.wikipedia.org/wiki/Generalized_LR_parser), which I
implemented in Emacs lisp.

The grammar file can be parsed by a new OpenToken feature, and it will
generate either Ada OpenToken source code, or Emacs lisp for a parser
table; that version of OpenToken is available on the Ada mode website
(not the OpenToken website http://stephe-leake.org/ada/opentoken.html).

However, OpenToken only has a (non-generalized) LALR parser. It would be
interesting to implement a generalized parser for Opentoken; let me know
if you'd like to use that and/or help implement it.

It might not be too hard to eliminate the conflicts so you can use the
OpenToken parser (or some other parser); I did not do that because it
seemed easier (not to say way more fun) to implement the parser :). In
addition, the grammar source is closer to the Ada LRM Annex P this way.

I've improved Ada mode a bit since the last post on the website; it's
quite usable now. The latest is in monotone; see the Ada mode website
for access info. It is time to post another release; maybe this weekend.

Also note that I fixed several bugs in OpenToken, so if it was not
working for you before, you should try it again.

-- 
-- Stephe