Appendix A
The Full Grammar
This appendix collects the complete context-free grammar of Nex in one place,
with the expression grammar given in its layered form so that operator
precedence and associativity (Section 2.6 ) are expressed directly by the
productions rather than assumed.
A.1 Programs, Modules, and Declarations
program ::= topitem * topitem ::= import | intern | classdec | uniondec | fundec | funsig | tydec | stmt import ::= import id (. id )* ⟨from string ⟩intern ::= intern id (/ id )* ⟨as id ⟩qid ::= id (/ id )* — class name, bare or qualified (Section 3.6.1 ) tydec ::= declare type id = ty ⟨where id : exp ⟩fundec ::= function id ⟨gen ⟩ ( ⟨params ⟩ ) ⟨: ty ⟩ ⟨note ⟩ ⟨require ⟩ do block ⟨ensure ⟩ ⟨rescue ⟩ end funsig ::= declare function id ⟨gen ⟩ ( ⟨params ⟩ ) ⟨: ty ⟩ ⟨note ⟩
A.2 Classes
classdec ::= ⟨sealed ⟩ ⟨deferred ⟩ class id ⟨gen ⟩ ⟨note ⟩ ⟨inherit ⟩ classbody ⟨invariant ⟩ end inherit ::= inherit parent (, parent )* parent ::= qid ⟨tyargs ⟩classbody ::= (featuresec | createsec )* uniondec ::= ⟨enum ⟩ union id ⟨gen ⟩ ⟨note ⟩ variant + end variant ::= id ⟨( params ) ⟩featuresec ::= ⟨private ⟩ feature member + createsec ::= create constructor + member ::= field | constant | method field ::= ⟨once ⟩ id : ty ⟨note ⟩ constant ::= id ⟨: ty ⟩ = exp ⟨note ⟩method ::= id ⟨( ⟨params ⟩ ) ⟩ ⟨: ty ⟩ ⟨alias ⟩ ⟨note ⟩ ⟨require ⟩ do block ⟨ensure ⟩ ⟨rescue ⟩ end | id ( ⟨params ⟩ ) ⟨: ty ⟩ ⟨alias ⟩ ⟨note ⟩ ⟨deferred ⟩alias ::= alias opsym opsym ::= "+" | "-" | "*" | "/" | "%" | "^" constructor ::= id ⟨( ⟨params ⟩ ) ⟩ ⟨require ⟩ do block ⟨ensure ⟩ ⟨rescue ⟩ end invariant ::= invariant assertion + params ::= param (, param )* param ::= id (, id )* ⟨: ty ⟩require ::= require assertion + ensure ::= ensure assertion + rescue ::= rescue block assertion ::= id : exp note ::= note string
A.3 Types and Generics
gen ::= [ genparam (, genparam )* ] genparam ::= ⟨? ⟩ id ⟨-> id ⟩ tyargs ::= [ ty (, ty )* ] ty ::= Integer | Real | Char | Boolean | String | ? ty | funty | qid ⟨tyargs ⟩funty ::= Function ⟨( ⟨funtyparams ⟩ ) ⟨: ty ⟩⟩funtyparams ::= funtyparam (, funtyparam )* funtyparam ::= id : ty | ty
A.4 Statements
block ::= stmt * stmt ::= id := exp | exp . id := exp | let id ⟨: ty ⟩ := exp | if exp then block (elseif exp then block )* ⟨else block ⟩ end | from block ⟨invariant ⟩ ⟨variant ⟩ until exp do block end | repeat exp do block end | across exp as id do block end | case exp of caseclause + ⟨else stmt ⟩ end | match exp of matchclause + ⟨else block ⟩ end | select selectclause + ⟨timeout ⟩ ⟨else block ⟩ end | do block ⟨rescue ⟩ end | with string do block end | raise exp | retry | assert (assertion + | exp ) | exp caseclause ::= scon (, scon )* then stmt matchclause ::= pattern ⟨as id ⟩ ⟨if exp ⟩ then stmt pattern ::= _ | qid ⟨tyargs ⟩ ⟨( fieldpat (, fieldpat )* ) ⟩fieldpat ::= _ | id | id : id | scon | pattern selectclause ::= when exp ⟨as id ⟩ then block timeout ::= timeout exp then block invariant ::= invariant assertion + variant ::= variant exp
A.5 Expressions (Layered)The following layering expresses the precedence and associativity of
Section 2.6 : each level refers to the next more tightly binding one, with the
binary operators of a level left-associative, ^ included
(Section 2.6 ).
exp ::= orexp orexp ::= andexp (or andexp )* andexp ::= eqexp (and eqexp )* eqexp ::= relexp ((= | /= | == | != ) relexp )* relexp ::= addexp ((< | <= | > | >= ) addexp )* addexp ::= mulexp ((+ | - ) mulexp )* mulexp ::= unexp ((* | / | % | ^ ) unexp )* unexp ::= - unexp | not unexp | ? postexp as id | postexp postexp ::= primary postfix * postfix ::= ⟨? ⟩ . id ⟨( ⟨args ⟩ ) ⟩ | ( ⟨args ⟩ ) primary ::= scon | id | this | super | result | ( exp ) | createexp | whenexp | fnexp | spawnexp | convexp | old primary | arraylit | maplit | setlit createexp ::= create qid ⟨tyargs ⟩ ⟨. id ⟨( ⟨args ⟩ ) ⟩⟩whenexp ::= when exp then exp else exp end fnexp ::= fn ⟨gen ⟩ ( ⟨params ⟩ ) ⟨: ty ⟩ do block end spawnexp ::= spawn do block end convexp ::= convert exp to id : ty args ::= exp (, exp )* arraylit ::= [ ⟨exp (, exp )* ⟩ ] maplit ::= { } | { exp : exp (, exp : exp )* } setlit ::= #{ ⟨exp (, exp )* ⟩ }
A.6 Lexical Grammar
id ::= (letter | _ ) (letter | _ | digit )* scon ::= int | byte | int16 | int32 | real | char | string | true | false | nil int ::= digit + | 0b bit + | 0o odigit + | 0x hdigit + byte ::= int u8 int16 ::= int i16 int32 ::= int i32 real ::= ⟨digit + ⟩ . digit + ⟨exp ⟩ char ::= # (char≠digit,ws | named | digit + )named ::= nul | space | newline | tab | return string ::= " … " | ' … ' comment ::= -- … end-of-line
A digit separator _ may occur between two digits of any numeric
literal. A byte , int16 or int32 is an
int with a suffix, forming a single token; its range is checked
(Section 2.2 ), and a leading - is absorbed into an
int16 or int32 . The reserved words of Section 2.1 are excluded from
id by the longest-match rule of Section 2.5 .