Reduce beanshell footprint

[sourceforge-issue-exporter]

I noticed that a large percentage of the beanshell footprint is devoted to the JavaCC lexer and grammar. The lex footprint could be reduced by using JFlex or another lex; the grammar footprint can be reduced by applying the optimization techniques described in the JavaCC doc (can't find the reference, says something about optimizing).

Reported by: fredthegorilla

Original Ticket: "beanshell/feature-requests/1":https://sourceforge.net/p/beanshell/feature-requests/1

[sourceforge-issue-exporter]

Logged In: YES user_id=18885

Although I am very happy with JavaCC, I was dissapointed that some of the promised enhancements never appeared... One of those was that the parser generator would get smarter with respect to some of the special optimizations possible with the lookaheads. The project seems to have turned its attention to the new metamata.com parser generator (which I have never evaluated). I did briefly look at the optimization docs for JavaCC but I didn't understand them very well at the time and wasn't sure that I could make the necessary changes to the grammar to support them. I am also hesitant to hand tune the output of the parser generator for maintenance reasons. As far as Flex I don't know what it would take to use an alternate lexer.

So, overall I agree and my sense is that the generated parser could be made smaller, but it will have to wait until someone tells me what to do or until I have time to look at it when things really settle down.

Thanks, Pat

Original comment by: patn

[sourceforge-issue-exporter]

• labels: 103273 -->
• milestone: 100100 -->

Original comment by: patn

[nickl-]

I tend to agree with Pat on this one, we are very much in bed with javacc lexer, and to change we would need an expert willing to port everything over to another framework. I don't think that is going to be as easily said than done.

[nickl-]

Closed replaced by #685