4 ms·

> Is there any formal name for these kinds of non-overloaded/simple syntaxes? If there is a formalism for those, why is it that "overloaded" syntaxes are prefer

by Wilduck 14y ago

> Is there any formal name for these kinds of non-overloaded/simple syntaxes? If there is a formalism for those, why is it that "overloaded" syntaxes are preferred most of the time, or why are they so prevalent? Someone might be able to say with more certainty, but I believe this is covered by the formalism of "Context Free Grammars" versus "Context Sensitive Grammars"[1]. That is, a language is context free if you don't need information from elsewhere in the code to determine the meaning of a symbol. There's a good discussion of what makes C context sensitive on Eli Bendersky's blog[2]. Edit: Thinking about this some more, I'm realizing that this has little to do with the overloading of symbols like block delimiters, since these can be tokenized and parsed by a context free grammar fairly easily in an "overloaded" fashion. Maybe what you're describing is related to the concept of "purely functional," or even some more vague notion of purity generally... [1] http://en.wikipedia.org/wiki/Context-free_grammar http://en.wikipedia.org/wiki/Context-free_grammar [2] http://eli.thegreenplace.net/2007/11/24/the-context-sensitivity-of-cs-grammar/#fn1 http://eli.thegreenplace.net/2007/11/24/the-context-sensitiv...

epidemian 14y ago
I don't think "context free" is the concept i was referring to. AFAIK, a grammar can be context free but still have these kind of "syntactic overloads" annoyances. E.g. in Python these expression will always be parsed the same no matter its context: "(foo(bar, baz) + (a,))"; it bothers me, however, that each one of those three pair of parentheses has a different syntactic meaning (one is for grouping, one is for function call, and the last one is for a tuple). Not ambiguous for the parser, but quite inconvenient for my brain's parser that i have distinguish between three possible different meanings for the same syntactic symbol.
- sly010 14y ago
  Such an unambiguous grammar would only have as many state transitions in its FSM than characters in the grammar, which would limit the language considerably, so you would have to raise the number of characters. Also ( means something different in a string, so in that sense evert programming language with strings are hard to parse locally. Also note that ambiguity formally means that more than one parse trees can represent the same string.

by Wilduck 14y ago

epidemian 14y ago

I don't think "context free" is the concept i was referring to. AFAIK, a grammar can be context free but still have these kind of "syntactic overloads" annoyances. E.g. in Python these expression will always be parsed the same no matter its context: "(foo(bar, baz) + (a,))"; it bothers me, however, that each one of those three pair of parentheses has a different syntactic meaning (one is for grouping, one is for function call, and the last one is for a tuple). Not ambiguous for the parser, but quite inconvenient for my brain's parser that i have distinguish between three possible different meanings for the same syntactic symbol.

sly010 14y ago
Such an unambiguous grammar would only have as many state transitions in its FSM than characters in the grammar, which would limit the language considerably, so you would have to raise the number of characters. Also ( means something different in a string, so in that sense evert programming language with strings are hard to parse locally. Also note that ambiguity formally means that more than one parse trees can represent the same string.