Re: ABNF Parser Generator

"Lowell Thomas" <lowell@coasttocoastresearch.com>
29 Nov 2005 16:13:21 -0500

          From comp.compilers

Related articles
ABNF Parser Generator lowell@coasttocoastresearch.com (Lowell Thomas) (2005-06-04)
Re: ABNF Parser Generator angray@beeb.net (Aaron Gray) (2005-06-16)
Re: ABNF Parser Generator ang-usenet@gmail.com (Aaron Gray) (2005-06-26)
Re: ABNF Parser Generator lowell@coasttocoastresearch.com (Lowell Thomas) (2005-11-29)
| List of all articles for this month |

From: "Lowell Thomas" <lowell@coasttocoastresearch.com>
Newsgroups: comp.compilers
Date: 29 Nov 2005 16:13:21 -0500
Organization: Compilers Central
Keywords: tools
Posted-Date: 29 Nov 2005 16:13:21 EST

ang-usenet@gmail.com (Aaron Gray) (2005-06-26) wrote:


>APG is not meta and does not seem to have its own bootstrap supplied.
>Its actions are event like which is fine for protocol recognition but it is
>not really geared up for normal gramatical and syntatic recognition of
>languages.
>Very interesting but not really upto the task I had in mind.
>Aaron


Aaron,


Thanks for your comments on APG - an ABNF Parser Generator. And especially
thanks for all the help and suggestions you have given me off line. But I
wanted to respond publicly to your comments to bring them up to date with the
latest version of APG, Version 3.1.


>APG is not meta


I'm still not quite sure what "meta" means in this context, but APG is a
parser identical in form to the parsers that it generates. Does that qualify
as "meta"? Is it important that it be "meta"?


> ... and does not seem to have its own bootstrap supplied.


True enough. No bootstrap has been released. Version 3.1 does give a
description of the bootstrap an how it works. The code itself is way to messy
for public consumption and until some compelling motivation comes along, it
will probably remain that way.


>Its actions are event like which is fine for protocol recognition but it is


Protocol recognition was the source of my original interest in parser
generation. And the fact that all IEFT RFCs are now described in ABNF was the
motivation to use that grammar format. But as I've gotten more familiar with
the needs of computer language recognition I have made a number of changes and
improvements. I've addressed the issues of syntactic and semantic predicates
and ambiguous grammars. Version 3.1 now comes with several sample applications
that deal with those issues.


>Very interesting but not really upto the task I had in mind.


Well, C++ is a challenge. But I think APG now provides a sufficient set of
tools for the attacking this problem. Whatever might still missing, I'm pretty
sure I can add. As time permits, I am pursuing the development of an ABNF
grammar for C++ and tinkering with the development of a C++ preprocessor and
eventually a complete front end. But learning as I go. I look forward to
suggestions you or anyone else in the comp.compilers community might have in
this regard.


Lowell Thomas
www.coasttocoastresearch.com



Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.