Let's assume Norman is really operating in the *SL context. There the meaning of 'form' can be resolved via Ryan's explanation, but with the understanding that these languages do not support macros. So read delivers something that the expander can turn into a core Racket feature or not. So the terms defined by the *SL grammars are the acceptable forms. -- Matthias
On May 16, 2014, at 1:00 PM, Sam Tobin-Hochstadt wrote: > I don't think this is really what Norman wants. Instead, I think he's > looking for a definition of `form` as it appears in various error > messages that mention "form". > > Sam > > > > On Fri, May 16, 2014 at 1:52 PM, Ryan Culpepper <ry...@ccs.neu.edu> wrote: >> If you refer to the "Syntax Model" section of the reference >> (http://docs.racket-lang.org/reference/syntax-model.html), you'll see that >> Racket syntax is defined by two passes, called "read" and "expand". >> >> The grammar you gave in your message is what I would call a "term"---it's >> anything that the read pass can turn into a syntax object. In other words, >> terms are what syntax objects represent. >> >> The word "form" generally means any of the non-terminals of the Racket >> language that the macro expander cares about, such as "expression forms" and >> "definition forms". There are also "module-level forms", which include >> require and provide. There's a list of these nonterminals in the subsection >> "Fully Expanded Programs" >> (http://docs.racket-lang.org/reference/syntax-model.html#%28part._fully-expanded%29). >> You can think of the syntax of Racket roughly as that grammar with "or a >> macro that eventually expands into one of the above" added to every >> production there---except the one for "formals". The "formals" non-terminal >> doesn't correspond to an expansion context, nor do "module-path" or >> "raw-require-spec"; thus, they are not considered forms. Neither is >> something like "cond-clause", for example; it's just an auxiliary >> non-terminal for the "cond" macro. >> >> To complicate matters, macros can also create new expansion contexts and >> thus new kinds of forms. For example, the class macro introduces "class-body >> forms", which include expressions and definitions as well as other syntaxes >> that are only significant immediately within class bodies, such as "init" >> and "abstract" forms. >> >> Ryan >> >> >> >> On 05/16/2014 12:28 PM, Norman Ramsey wrote: >>> >>> Where would I find out, preferably in the Racket documentation, >>> the definition of what a 'form' is? I'm thinking along these lines: >>> >>> A form is one of: >>> - a symbol >>> - a numeric literal >>> - a string literal >>> - ... possibly other literals? ... >>> - a sequence of forms wrapped in one pair of parentheses >>> >>> Is this the right idea? Or is the word 'form' reserved only for >>> things wrapped in parentheses? >>> >>> >>> Norman >>> >>> >>> P.S. In the Racket documentation, I found the section on 'Syntactic >>> Forms', but I haven't found a definition of what a form is, and I >>> haven't found the word 'form' in the index. >>> >>> >>> ____________________ >>> Racket Users list: >>> http://lists.racket-lang.org/users >>> >> >> ____________________ >> Racket Users list: >> http://lists.racket-lang.org/users > ____________________ > Racket Users list: > http://lists.racket-lang.org/users ____________________ Racket Users list: http://lists.racket-lang.org/users