Feature Request: Inverse for syntax exclusion

[abort]

During development of a javascript grammar for rascal, exclusion is widely used to only search for 'specific' types of statements. For now, we are getting huge lines like;
Statement!c!d!e!f!g!h!i!j!k.... etcetera, for statements to be just of type or label label/type 'a' and 'b'. In that case it would be more efficient (but more importantly, easier to understand and maintain) to use a form like Statement#a#b or Statement/a/b, where we only expect to match on a and b. This would be the inverse of the exclusion.

[SanderBenschop]

The following is an example from the JavaScript grammar we are working on. We want a few specific types of statements to match and exclude the rest, without altering syntax or creating redundant depth/overhead in our tree.

Statement!variableSemi!expressionSemi!returnExp!throwExp!returnNoExp!throwNoExp!continueLabel!continueNoLabel!breakLabel!breakNoLabel!empty!expressionLoose!emptyBlockEnd!continueLabelNoSemiBlockEnd!breakLabelNoSemiBlockEnd!continueNoLabelNoSemiBlockEnd!breakNoLabelNoSemiBlockEnd!returnExpNoSemiBlockEnd!returnNoExpNoSemiBlockEnd!throwExpNoSemiBlockEnd!throwNoExpNoSemiBlockEnd!expressionBlockEnd!block!ifThen!ifThenElse!doWhile!whileDo!forDo!forIn!tryBlock!switchCase

[grammarware]

If some small part of a nonterminal is needed relatively often, it could make sense to promote it to a different nonterminal, which would eliminate the problem entirely. Exclusion is a stretch for small hacks.

[jurgenvinju]

or an alias could also be nice. I think Sander does not want to change the
shape of the parse forest while restricting to only a few alternatives. I
agree that that is wat a non-terminal is also for. It's interesting :-)

On Thu, Mar 13, 2014 at 1:42 PM, Vadim Zaytsev notifications@github.comwrote:

If some small part of a nonterminal is needed relatively often, it could
make sense to promote it to a different nonterminal, which would eliminate
the problem entirely. Exclusion is a stretch for small hacks.

Reply to this email directly or view it on GitHubhttps://github.com//issues/518#issuecomment-37528093
.

Jurgen Vinju

• Centrum Wiskunde & Informatica - SWAT

• INRIA Lille - ATEAMS

• Universiteit van Amsterdam

  www: http://jurgen.vinju.org, http://www.rascal-mpl.org,
  http://twitter.com/jurgenvinju
  skype: jurgen.vinju

[afroozeh]

I faced the same problem Sander has when I was working with the OCaml grammar, so I think it's good to have this feature inn Rascal.

Semantically, adding a nonterminal makes sense, but in addition to changing the parse tree, introducing a new nonterminal will also affect the performance of parsing: GSS calls to the original nonterminal and SPPF nodes with the same label as the original nonterminal are not shared.

I think the alias idea is nice. I was thinking of something like this:

syntax E =  E + E1
       | E - E2
       | - E 
       | a

alias E1 from E with { E - E, E + E}

alias E2 from E without {  - E }

or just simply

alias E2 from E = 
        E + E
        | E - E
        | a