[antlr-interest] Dynamic token generation

[Jan van der Ven]

Dear Antlr Community,


First of all, let me tell you that our SourceForge project is still 
using Antlr 2.7. We are not old-fashioned, we just do not have the time 
to cope with all the new versions of the splendid tools we are working with.

Having said that, my problem as described below, might force an upgrade 
and if required we will do so.

We are extending the QuantumDB eclipse plugin 
(http://sourceforge.net/projects/quantum) with grammar recognition 
capabilities. You can find the base grammar file here: 
http://quantum.cvs.sourceforge.net/quantum/quantum-plugin/src/com/quantum/core/sql/grammar/DmlSQL2.g?view=log. 
It is an adaptation of Mr Vnuk's DmlSQL2.g file. It is full of code. It 
handles only SQL-92. Any feedback on the file as it is will be 
appreciated, but the problem at hand is the following:

- Each database has its own set of keywords
- Each database has its own set of functions

Quantum uses a JDBC driver to connect to the specific database. Most 
drivers report a list of keywords and functions. I would like these to 
appear as tokens in the grammar. This means that a rule would refer to a 
token as a KEYWORD, but the KEYWORD should be defined at runtime. Is 
this possible?

Kind regards,


Jan