sdd commented on code in PR #320: URL: https://github.com/apache/iceberg-rust/pull/320#discussion_r1554438555
########## crates/iceberg/src/expr/visitors/bound_predicate_visitor.rs: ########## @@ -0,0 +1,366 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +use crate::expr::{BoundPredicate, BoundReference, PredicateOperator}; +use crate::spec::Datum; +use crate::Result; +use fnv::FnvHashSet; + +pub trait BoundPredicateVisitor { + type T; + + fn always_true(&mut self) -> Result<Self::T>; + fn always_false(&mut self) -> Result<Self::T>; + + fn and(&mut self, lhs: Self::T, rhs: Self::T) -> Result<Self::T>; + fn or(&mut self, lhs: Self::T, rhs: Self::T) -> Result<Self::T>; + fn not(&mut self, inner: Self::T) -> Result<Self::T>; + + fn is_null(&mut self, reference: &BoundReference) -> Result<Self::T>; Review Comment: I don't see how this will work with just `visitor_op`. The proposed function signature `fn visitor_op(&mut self, op: &PredicateOperator, reference: &BoundReference)` would work for a unary operator. A binary operator would need `fn visitor_op(&mut self, op: &PredicateOperator, reference: &BoundReference, literal: &Datum)`, and a set operator would need `fn visitor_op(&mut self, op: &PredicateOperator, reference: &BoundReference, literals: &FnvHashSet<Datum>)`. To resolve this, you'd need the visitor to have `fn unary_op(...)`, `fn binary_op(...)`, and `fn set_op(...)`. Is that what we want? Personally I think this is another example of where the current type hierarchy feels awkward - there are other places where we bump into these type hierarchy mismatches, such as needing a default leg in a match on operator to handle operators that will never show up, such as binary ops in a unary expression. Something more like this feels better to me: ```rust pub enum Predicate { // By not having Unary / Binary / Set here, and removing // that layer in favour of having the ops themselves // as part of this enum, there is no need for code that // raises errors when the wrong type of `Op` is used, since those // issues are prevented by the type system at compile time // Unary ops now only contain one param, the term, and so can be // implemented more simply IsNull(Reference), // ... other Unary ops here, // Binary / Set ops ditch the `BinaryExpression`, with the literal // and reference being available directly GreaterThan { literal: Datum, reference: Reference, }, // ... other Binary ops here, In { literals: HashSet<Datum>, reference: Reference, }, // ... other Set ops here } ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
