The case-insensitive attribute on a template production is supposed to make all keywords case-insensitive (source, source). With the new Java parse table generator (sdf2table: java) in Spoofax Eclipse 2.4.0.20171026-103626-master, the AST is wrong. For example, given the following (simplified Pascal) language:

module foo

context-free start-symbols Start

context-free syntax
  
  Start.Program = <
    <Programheading> 
  >

  Programheading.ProgramHeading = <
    program <Identifier> (<{Identifier ", "}*>);
  > {case-insensitive}

lexical syntax // Variable and Identifier Categories

  Identifier = [a-zA-Z][a-zA-Z0-9\_]*
  
lexical restrictions

  Identifier -/- [a-zA-Z0-9\_]

lexical syntax // layout
  
  LAYOUT           = [\ \t\n\r] // whitespace
  
  LAYOUT           = Comment
  
  Comment          = "{" InsideComment* "}" 
  InsideComment    = ~[\}] 
  InsideComment    = Comment // nested comment, not in standard
  
  Comment          = "(*" InsideParComment* "*)" 
  InsideParComment = ~[\*]
  InsideParComment = Asterisk
  Asterisk         = "*"

lexical restrictions

  Asterisk -/- [\)]
  
context-free restrictions

  LAYOUT? -/- [\ \t\n\r\{]

parsing the following program:

program x ();

yields the following AST:

Program(
  ProgramHeading(
    (112, 114, 111, 103, 114, 97, 109)
  , "x"
  , 40
  , []
  , 41
  , 59
  )
)

Instead, one would expect:

Program(
  ProgramHeading(
    "x"
  , []
  )
)

Indeed, with the C parse table generator (sdf2table: c), we get the latter AST.

Submitted by Martijn on 16 November 2017 at 22:08

On 23 February 2018 at 14:58 Eduardo Amorim commented:

This issue should be fixed in the latest/nightly version of Spoofax.

Log in to post comments