Re: Free Ada parser

[Craig Carey <research@ijs.co.nz>]

On Sun, 14 Dec 2003 11:57:51 +1300, Craig Carey wrote:


>I uploaded a package that has merged Ada [A]Yacc parser lines.

I.e. SCATC and NewP2Ada got merged. If there is a version with fixess not
in those 2 then I would like to know of that.


>... has a demo program that makes [*use*] of the Adagoop program.
>   http://www.ijs.co.nz/code/ada95_adagoop_parser.zip


I get some of these wording errors out

>Two other document[*ed*] uses are [...]

I use Adagoopp and people at the USAFA use it. 

-------------

On 13 Dec 2003 17:46:17 +0100, Pascal Obry wrote:
...
>And if this version does not work you I have ported a version of aflex/ayacc
>(long time ago). You can download this from my homepage.
...
http://perso.wanadoo.fr/pascal.obry/archive/flexyacc.tar.gz
  "aflex/ayacc (9/3/1997)"

I guess that fixes in that gzip file are in NewP2Ada and my code.

Also it puts Ada code in case statement alternatives.

 * With case statement alternatives, the case alternative that sets a
   mode for a subexpression may be different from the case alternative
   that resets the mode.

   I don't know if changing the grammar can always provide a solution.

 * when Adagoop is used, probably recursion is used. So every mode that
   is set can be reset or unset. 

   E.g.  the quote in "x or '(0<a)" prevents the replacing of 'a' with
   its value.

Ada might have to be avoided if Yacc is not good enough, and debugging
Yacc is an area that could lead to a rejection of the Ada Yacc parser.

If YAcc was replaced with the moder advanced ANTLR C++ compiler, a
likely main problem could remain:

   http://www.bearcave.com/software/antlr/antlr_examples.html

  "I developed a "tiny C" grammar for ANTLR annotated to
   produce trees to better understand what a real ANTLR
   language front end would be like. One of the things that I
   learned with this grammar is that debugging ANTLR grammars
   is not that much easier than, say, debugging a YACC grammar.
   There was a mistake in the production describing the for-
   loop. ANTLR gave a non-determinism warning for the unary
   minus production. I spent hours pouring over my expression
   grammar containing the unary minus, but I could find no
   error in it."




Craig Carey