Talk:Recursive descent parser generator

From Rosetta Code

Input specification

The main thing that needs to be fixed before adding any examples is the input specification. I want to make it detailed enough to be at least slightly usable (so more than just a recognizer) but I'm having trouble coming up with a clear way of expressing both the rule and the extra code generated when the rule is applied, especially taking in to account all the different languages that you might want to output with or to. Here's one attempt:

header code to output

!! start -> var expr
more code

!! expr -> '+' expr

Is there any way to simplify this? (I can't remember, can you do everything that needs to be done with a header and just post-traversal output?)

The example I want to use is to take an arithmetic expression and turn it in to a series of single operations in the right order. So:

(one + two) * three + four * five

Becomes something like:

_a = one + two
_b = _a * three
_c = four * five
_d = _b + _c

Oh yeah, if the task specification says to output the parser in your language of choice, the example will have to be rewritten for each language too, unless there's some way around that.

MagiMaster (talk) 06:10, 2 April 2014 (UTC)

I think I'd give some input samples and describe the characteristics of the output. Then I'd provide a sample implementation. If people have questions or whatever that will show up soon enough. --Rdm (talk) 06:53, 2 April 2014 (UTC)

I want to make it detailed enough to be at least slightly usable
The above can work against making a good RC task sometimes. You really don't want to make things too long. You also want to make a task in which the examples generated are comparable. --Paddy3118 (talk) 07:12, 2 April 2014 (UTC)
I think if it can handle the above example, it'd be enough. MagiMaster (talk) 16:49, 2 April 2014 (UTC)
For what it's worth: the implementation I provided handles the example, but uses the same precedence for multiplication and addition. I would not have had to do it that way, but was hoping for more examples before I added the line(s) precedence handling would require. (The trick is that the parser resolves from right to left, so it's result happens to coincide with the result you would get from a left to right parser with multiplication having a tighter precedence than addition or subtraction.) --Rdm (talk) 18:38, 13 December 2019 (UTC)