Make generated parsers smaller

[dmajda]

Generated parsers are too big:

dmajda@inuit:~/Programování/Projekty/PEG.js/pegjs$ bin/pegjs examples/arithmetics.pegjs 
dmajda@inuit:~/Programování/Projekty/PEG.js/pegjs$ bin/pegjs examples/css.pegjs 
dmajda@inuit:~/Programování/Projekty/PEG.js/pegjs$ bin/pegjs examples/javascript.pegjs 
dmajda@inuit:~/Programování/Projekty/PEG.js/pegjs$ bin/pegjs examples/json.pegjs 
dmajda@inuit:~/Programování/Projekty/PEG.js/pegjs$ wc -c examples/*.js
  13327 examples/arithmetics.js
 429010 examples/css.js
 610664 examples/javascript.js
  54518 examples/json.js
1107519 celkem
dmajda@inuit:~/Programování/Projekty/PEG.js/pegjs$ wc -c src/parser.js 
103838 src/parser.js

They should be much smaller, ideally without sacrificing performance.

[ceymard]

I am not sure it would be that useful to make generate smaller parsers an important issue : Closure Compiler or other minifiers do a good job at that.

Performance ought probably to be the most important aspect of pegjs, imho.

[dmajda]

@pwipwi: For some people the most important thing is performance, for some size. It totally depends on usage (web/Node.js, small inputs vs. big inputs, simple grammars vs. complex grammars, etc.). Minifiers can help of course, but not everybody uses them and they also can't compress the code structurally (much).

The size improvements shouldn't primarily be things like removing whitespace and making all identifiers just one letter long (that's the job for minifiers). They should be mostly removing useless/duplicate code and making it simpler where possible.

I am also thinking about a state-machine/bytecode-based approach instead of generating JavaScript code directly — that would probably help a lot here.

[ceymard]

Agreed.

[dmajda]

Fix to reset parser positions when action returns |null| made generated parsers somewhat larger, but today's variable handling rewrite made them almost 9% smaller.

[dmajda]

This was fixed by the code generator rewrite.

More optimizations are possible but they should be tracked separately elsewhere.