Grammar railroad diagram

[mingodad]

Using the syntax described here https://dusa.rocks/docs/language/syntax/ after small fixes (<builtin-identifier> probably is <builtin>) I've included it here https://mingodad.github.io/parsertl-playground/playground/ (an Yacc/Lex compatible online interpreter/editor, select Gusa parser from Examples then click Parse to see a parser tree for the content in Input source) and from there generated the EBNF understood by https://github.com/GuntherRademacher/rr that generate a nice navigable railroad diagram that can be viewed online (see instructions at the top bellow).

//
// EBNF to be viewd at
//    (IPV6) https://www.bottlecaps.de/rr/ui
//    (IPV4) https://rr.red-dove.com/ui
//
// Copy and paste this at one of the urls shown above in the 'Edit Grammar' tab
// then click the 'View Diagram' tab.
//

program::=
	  program declaration
	| /*%empty*/

declaration::=
	  "#builtin" builtin identifier dot_opt
	| "#lazy" identifier dot_opt
	| "#demand" premises '.'
	| "#forbid" premises '.'
	| conclusion '.'
	| conclusion ":-" premises '.'

dot_opt::=
	  /*%empty*/
	| '.'

premises::=
	  premise
	| premises ',' premise

premise::=
	  term oper term
	| attribute
	| attribute "is" term

oper::=
	  "=="
	| "!="
	| ">="
	| '>'
	| "<="
	| '<'

conclusion::=
	  attribute
	| attribute "is" term_or_choices
	| attribute "is?" term_or_choices

term_or_choices::=
	  term
	| '{' choice_options '}'

choice_options::=
	  term
	| choice_options ',' term

attribute::=
	  identifier
	| identifier arguments

arguments::=
	  atomic_term
	| arguments atomic_term

atomic_term::=
	  wildcard
	| variable
	| string_literal
	| int_literal
	| identifier
	| builtin
	| '(' term ')'

term::=
	  atomic_term
	| identifier arguments
	| builtin arguments

builtin::=
	  "BOOLEAN_TRUE"
	| "BOOLEAN_FALSE"
	| "NAT_ZERO"
	| "NAT_SUCC"
	| "INT_PLUS"
	| "INT_MINUS"
	| "INT_TIMES"
	| "STRING_CONCAT"

//Tokens

COMMENT ::= "#"[ #x09].*
variable ::= [A-Z][a-zA-Z0-9_]*
wildcard::= '_'[a-zA-Z0-9_]*	 // and represents variable names that you wish to be ignored.
identifier ::= [a-z][a-zA-Z0-9_]*
string_literal ::= '"'[^"]+'"'	 //is a regular string constant with no escape characters: two double quotes " surrounding any ASCII character in the range 32-126 except for " and \.
int_literal ::= '0'|'-'?[1-9][0-9]*

[mingodad]

I think that calling the grammar in https://dusa.rocks/docs/language/syntax/ as Context-free grammar it's no true because you have <declaration> ::= "#builtin" <builtin> <identifier> [ "." that makes it a Context-sensitive grammar.

[spencerflem]

<builtin-identifier> probably is <builtin>

i've been puzzling this for a bit. I think this is wrong,

try code like:

outcome "rock" "scissors".
outcome "scissors" "paper".
outcome "paper" "rock".

player "player1".
player "player2".
move Move :- outcome Move _.

plays P N is? Move :- round N, player P, move Move.

round 1.
round (INT_PLUS Round 1) :-
  plays "player1" Round is Move,
  plays "player2" Round is Move.

if <builtin-identifier> was a mystyped <builtin> then INT_PLUS would be valid here, but it is not.

I believe that <builtin-identifier> is either just a normal <identifier> and the distinction should be removed, or else <builtin-indentifier> should be added to Tokens and "#builtin" <builtin> <identifier> [ "." ] should be "#builtin" <builtin> <builtin-identifier> [ "." ]

[mingodad]

It seems that #builtin is like typedefin C/C++ and so as I said before the grammar is context sensitive.
Edited: case sensitive -> context sensitive.

[robsimmons]

I think the best way to fix the grammar here is to just replace <builtin-identifier> with <identifier> — the initial syntax parse doesn't distinguish identifiers (a previous iteration probably did, and this is just an artifact).

[robsimmons]

@mingodad with #103 merged the syntax spec is (correctly) no longer context sensitive — the parser hasn't been relying on the context of previous builtin declarations for awhile