Equation evaluator

> "rush" <[hidden email]> wrote in message
> news:al9mn2$1oq6$[hidden email]...
>
>>"John Brant" <[hidden email]> wrote in message
>>news:jgOd9.1346$Jo.226@rwcrnsc53...
>>
>>
>>>Without the updated compiler, Dolphin doesn't compile literal byte
>>
> arrays
>
>>>inside of literal arrays (e.g., "#(#[])"). It is possible to get around
>>
>>this
>>
>>>bug by using compile time constants (e.g., "#(##(#[]))"), but I didn't
>>
>>want
>>
>>>to change SmaCC's code that much. You can change the code if you want.
>>
> You
>
>>>need to change all of the #transitionTable methods to use compile time
>>>constants (search/replace "#[" -> "##(#[" and "]" -> "])"). You also
>>
> need
>
>>to
>>
>>>change the SmaCCGrammarCompiler>>compileTransitionTable method to output
>>>'##(#[' and '])' instead of '#[' and ']' respectively.
>>
>>is it only in the code of SmallCC or also in the generated code?
>
>
> The #transitionTable methods are all generated methods. However, if you also
> change the #compileTransitionTable method, then it would generate the code
> that used the compile time constants. Of course, with the updated compiler
> that Blair posted, you shouldn't need to change anything.
>
>
>>Can SmallCC
>>code be used in commercial programs?
>
>
> Yes, we only request that if you make some grammar that could be useful for
> others, you post it. A little acknowledgement never hurts either :).
>
> BTW, if anyone has any problems, please report them. The development is done
> under VW and ported to VA and Dolphin using a script that apply several
> rewrite rules. We could be missing some necessary transformations in our
> script.
>
>
> John Brant
>
>

> I say, "I think", because I am now stuck at the next hurdle.  Part of
> the app I am attempting wull let the user enter an arbitrary (but
> simple, at least at first) equation such as "(inputA + inputB) / inputC
> = output".  My knowledge of parsing and evaluating is woefully
> inadequate.
>
> I think I need to generate some kind of tree into which I can substitute
> values and evaluate it.  Does anyone know of some good resources on this
> sort of thing?  (I suspect if I had done a computing degree it would
> have been covered in some depth :-( )

> "Tom Ayerst" <[hidden email]> wrote in message
> news:[hidden email]...
>
>
>>I say, "I think", because I am now stuck at the next hurdle.  Part of
>>the app I am attempting wull let the user enter an arbitrary (but
>>simple, at least at first) equation such as "(inputA + inputB) / inputC
>>= output".  My knowledge of parsing and evaluating is woefully
>>inadequate.
>>
>>I think I need to generate some kind of tree into which I can substitute
>>values and evaluate it.  Does anyone know of some good resources on this
>>sort of thing?  (I suspect if I had done a computing degree it would
>>have been covered in some depth :-( )
>
>
> You don't have to create a parse tree, but it may be better. Will the same
> expression be evaluated multiple times? If so, then it is probably better to
> create a parse tree that can be interpreted. You should look at the
> Interpreter pattern from the Design Patterns book (or the Design Patterns
> Smalltalk Companion book).
>
> If you are only executing the expression once (e.g., an initialization
> expression), then you can get by without needing a tree to interpret. For
> example, we can add variables to our simple calculator example
> (http://www.refactory.com/Software/SmaCC/Tutorial.html) by doing the
> following:
>
> 1) Add an instance "values" to the CalculatorParser class
> 2) Add a setter method for the "values" variable:
>     CalculatorParser>>values: aDictionary
>         values := aDictionary
> 3) Add a new method that parses an expression with some variables:
>     CalculatorParser class>>parse: aString withValues: aDictionary
>         | parser |
>         parser := self on: (ReadStream on: aString).
>         parser values: aDictionary.
>         ^parser parse
> 4) Add "<variable> : [a-zA-Z_] \w* ;" to the calculator scanner's definition
> 5) Add a new rule for the Number production in the parser's definition:
>     | <variable> 'variable'
>         {values
>             at: variable value
>             ifAbsent:
>                 [self reportErrorMessage: variable value, ' is undeclared']}
> 6) Now compile the parser and run a test like:
>     | variables |
>     variables := Dictionary new.
>     variables
>         at: 'one' put: 1;
>         at: 'var' put: 214.
>     CalculatorParser parse: 'var / 2 + one' withValues: variables
> You should get 108. Also, if you use a name that isn't in your dictionary,
> you should get an error.
>
>
> John Brant
>
>