greg/mentat
1
0
Fork 0
Code Issues 294 Pull requests Packages Projects Releases 9 Wiki Activity

(#362) Part 1: if a variable's type becomes known, don't extract it. r=nalexander

Browse source 
This is necessary because we process patterns sequentially; a later
pattern might tell us the type of a variable (e.g., by having a
constant attribute), at which point we can do less work.
This commit is contained in:
Richard Newman 2017-03-06 20:16:14 -08:00
parent 1961815acd
commit 8935d6a8a5
1 changed files with 14 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

14
query-algebrizer/src/cc.rs
View file

@ -244,6 +244,8 @@ impl ConjoiningClauses {
!self.known_types.contains_key(&var) &&
!self.extracted_types.contains_key(&var);
// If we subsequently find out its type, we'll remove this later -- see
// the removal in `constrain_var_to_type`.
if needs_type_extraction {
self.extracted_types.insert(var.clone(), alias.for_type_tag());
}
@ -266,6 +268,18 @@ impl ConjoiningClauses {
/// Returns `false` if it's impossible for this type to apply (because there's a conflicting
/// type already known).
fn constrain_var_to_type(&mut self, variable: Variable, this_type: ValueType) -> bool {
// If this variable now has a known attribute, we can unhook extracted types for
// any other instances of that variable.
// For example, given
//
// ```edn
// [:find ?v :where [?x ?a ?v] [?y :foo/int ?v]]
// ```
//
// we will initially choose to extract the type tag for `?v`, but on encountering
// the second pattern we can avoid that.
self.extracted_types.remove(&variable);
// Is there an existing binding for this variable?
let types_entry = self.known_types.entry(variable);
match types_entry {