ADQL grammar validation

[Dave Morris]

Hi DAL,

Progress on developing a method for validating the BNF grammar for ADQL 
has been stalled for a while.

Following on from discussions at the ASTERICS meeting in Strasbourg, I 
have been reading up on language grammars and BNF. Based on what I 
learned I've put together a proposal for how to go about creating a 
machine readable language grammar for ADQL.

Short form is :- rather than trying to patch fix the existing BNF 
grammar we create a new grammar from scratch, adding each of the 
language rules and validating them one at a time.

More detail is available on the project wiki.
https://github.com/ivoa/lyonetia/wiki/BNF-grammar-and-validation

We have found a tool that can validate a BNF (or EBNF) grammar against a 
set of examples:
http://bnfparser2.sourceforge.net/

And we have set up a prototype system in a Docker container for 
developing and validating a grammar for ADQL:
https://github.com/ivoa/lyonetia/tree/master/src/docker/bnfparser

--------

I'd be interested to hear what you think.

Is this a realistic approach to creating a machine readable grammar for 
ADQL ?

Do you have an alternative method you would like to suggest ?

At this stage we would welcome any and all suggestions on how best to 
create a machine readable grammar and how to implement the validation 
process.

Thanks,
Dave

--------
Dave Morris
Research Software Engineer
Wide Field Astronomy Unit
Institute for Astronomy
University of Edinburgh
--------