Precedence operators to resolve S/R conflicts

[kivikakk]

This isn't a feature request, more feeling out: would you be interested in a PR that added precedence operators to automatically solve shift–reduce conflicts? (per bison's %left / %right / %nonassoc) If so, I might give it a shot. (No guarantees I'm actually capable of it!)

[goffrie]

There's already some support for this, though undocumented, haha. Look for
"overriding" and "no_reduce".

On Oct 18, 2016 19:24, "Yuki Izumi" notifications@github.com wrote:

This isn't a feature request, more feeling out: would you be interested in
a PR that added precedence operators to automatically solve shift–reduce
conflicts? (per bison's %left / %right / %nonassoc) If so, I might give
it a shot. (No guarantees I'm actually capable of it!)

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#12, or mute the thread
https://github.com/notifications/unsubscribe-auth/ABLtTrE3vbNDGrmAc7aeBI-vJQ-W2vcgks5q1X9igaJpZM4KafmF
.

[goffrie]

See https://github.com/rudi-c/cs444-java-compiler-in-rust/blob/master/ast/src/parser.rs for an old example of how I used them. If you want to improve the interface or make them more general, or whatever, that would be great!

[kivikakk]

Great, will look at this tonight! Thanks :)

[Addisonbean]

Sorry if this is the wrong place for this, but I am getting a shift reduce conflict, and am unable to figure out how to use these macros to fix it. Here is the code:

expr: Expr {
    // #[overriding]
    Ident(var) Equals expr[rhs] => Expr {
        span: span!(),
        node: Expr_::Assign(var, Box::new(rhs)),
    },
    expr[e] LParen RParen => Expr {
        span: span!(),
        node: Expr_::FuncCall(e),
    },
    term[t] => t,
}

That second line, the one commented out, was my attempt at trying to solve the error, but that did nothing to fix it. Do you have any idea how I'd use these macros to fix this problem?

Here is the error:

error: shift-reduce conflict:
state: expr -> Ident Equals expr •
       expr -> expr • LParen RParen
token: LParen

Thank you for your time!

[goffrie]

So I believe you can fix this by using #[no_reduce(LParen)], but I don't recommend that as a proper solution. The normal way to fix this kind of issue in an LR grammar is to introduce precedence levels for different kinds of expressions (in this case, you'd want the function-call expression to have higher precedence than the assignment-expression). I don't know of a good reference to point you to off-hand, but any classical compiler textbook should cover this kind of thing. Hope this helps!

[Addisonbean]

The #[no_reduce(LParen)] worked great, and then I was able to get it working with out that by looking further into that java compiler link your shared here. Thank you!