[antlr-interest] "An Introduction to ANTLR" presentation slides

[Andy Tripp]

Loring Craymer wrote:
> *As the noted logician Charles Lutwidge Dodgson wrote
> March Hare:* …Then you should say what you mean.
> *Alice:* I do; at least - at least I mean what I say -- that's the 
> same thing, you know.
> *Hatter:* Not the same thing a bit! Why, you might just as well say 
> that, 'I see what I eat' is the same as 'I eat what I see'!
> *March Hare:* You might just as well say, that "I like what I get" is 
> the same thing as "I get what I like"!
> *The Dormouse:* You might just as well say, that "I breathe when I 
> sleep" is the same thing as "I sleep when I breathe"!
>
> The same lexer that got you 'cat' will get you 'chat'. 
And the same person that got you "A Cat is a Feline" will get you "A 
chat is a conversation".
To the lexer, "cat" is the meaning of some input.
> What meaning would you give that?
> Now suppose the sentence was "Viva la chat!".  Would your 
> interpretation change?
Not if I'm a lexer, no.
If I'm a parser, no.
If I'm a treealker, no.
In all three cases, "char" is still a word - some  output token passed 
from a lexer to a parser.

As a person, of course. We have an entirely different architecture.

This example is a good illustration of why I keep harping on each phase 
(lexer, parser, treewalker)
having it's own idea of what's valid input. And it doesn't make much 
sense to me to
refer to all these different types of input as all "syntax". Suppose we 
have two lexers (one for
ASCII only, one that handles unicode". And suppose we have two parsers, 
one for English
sentences and another for French. And we have a single treewalker that 
accepts
AST input from either parser. Most people, including compiler writers, 
would look
at a particular sentence and when asked "Is that correct syntax?", would 
give just one
answer, not five. That says to me that "syntax" means something specific 
to them, and
it doesn't mean "the form of any sort of input".
>
> Lexers understand the character syntax from which words are built, not 
> the semantics of the words.
Well, no program really "understands" syntax or semantics.
It's just software manipulating symbols.

So, I'd reword that to say lexers *accept* certain input.
Here, I think sometimes people do talk about a lexer accepting certain 
*syntax*, but more often,
they'll say it accepts a certain "set of characters" or something like that.

So, take the "understanding" parts out, and what do you have?
Any software just takes input and produces output.
I would call that process "assigning meaning". It certainly can be done 
either with or without
"semantic analysis".
>
> --Loring

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.antlr.org/pipermail/antlr-interest/attachments/20080303/d41f1939/attachment.html